Download Climate Regulation and the Limits of Cost

University of Chicago Law School
Chicago Unbound
Public Law and Legal Theory Working Papers
Working Papers
2010
Climate Regulation and the Limits of Cost-Benefit
Analysis
Eric A. Posner
Jonathan Masur
Follow this and additional works at: http://chicagounbound.uchicago.edu/
public_law_and_legal_theory
Part of the Law Commons
Recommended Citation
Eric Posner & Jonathan Masur, "Climate Regulation and the Limits of Cost-Benefit Analysis," University of Chicago Public Law &
Legal Theory Working Paper, No. 315 (2010).
This Working Paper is brought to you for free and open access by the Working Papers at Chicago Unbound. It has been accepted for inclusion in Public
Law and Legal Theory Working Papers by an authorized administrator of Chicago Unbound. For more information, please contact
[email protected].
CHICAGO JOHN M. OLIN LAW & ECONOMICS WORKING PAPER NO. 525 (2D SERIES) PUBLIC LAW AND LEGAL THEORY WORKING PAPER NO. 315 CLIMATE REGULATION AND THE LIMITS OF COST‐BENEFIT ANALYSIS Jonathan S. Masur and Eric A. Posner THE LAW SCHOOL THE UNIVERSITY OF CHICAGO August 2010 This paper can be downloaded without charge at the John M. Olin Program in Law and Economics Working Paper Series: http://www.law.uchicago.edu/Lawecon/index.html and at the Public Law and Legal Theory Working Paper Series: http://www.law.uchicago.edu/academics/publiclaw/index.html and The Social Science Research Network Electronic Paper Collection. Climate Regulation and the Limits of Cost-Benefit Analysis
Jonathan S. Masur and Eric A. Posner1
July 29, 2010
Introduction
There is by now a broad consensus that global warming threatens significant harm to the
welfare of people across the world and that national governments should take steps to curb
warming and alleviate the harm caused by climate change.2 After dragging its heels, the United
States has joined most other countries in recognizing that action against climate change is
needed. Nonetheless, the United States has not ratified any international climate change treaty.
Congress has considered, but not passed, greenhouse gas regulation.3 And the EPA has not yet
proposed broad carbon dioxide regulation,4 despite having been authorized to do so by the
Supreme Court.5 To all appearances, the federal government is at a point of stasis—deliberating,
but not acting, on the issue of climate change.
Yet this appearance is misleading: in fact, the American regulatory state has lurched into
gear. Over the past two years, several federal agencies, most notably the Departments of Energy
and Transportation, have issued fourteen regulations that take into account the effect of industrial
activities and products on the global climate. The regulatory activity so far involves relatively
small-scale projects, for example, energy standards for appliances like air conditioners and gas
ranges. But these early efforts by agencies to engage in climate regulation reveal the numerous
challenges they face, and how they are likely to meet them when they turn to larger scale
regulation involving power plants, motor vehicle regulation, and manufacturing. The regulatory
activity so far has already set precedents on which future regulation will rest, yet despite the
potentially momentous consequences, it has received no comment in the law review literature.
1
University of Chicago Law School. Thanks to William Hubbard, Saul Levmore, Mike Schill, Naomi Schoenbaum,
David Weisbach, and an audience at the University of Chicago Law School for helpful comments, and to Paras
Bhayani and James Kraehenbuehl for research assistance.
2
See, e.g., William Nordhaus, A Question of Balance: Weight the Options on Global Warming Policies (2008);
Nicholas Stern, The Economics Of Climate Change (2007); Richard B. Stewart & Jonathan B. Wiener,
Reconstructing Climate Policy: Beyond Kyoto (2003); Cass R. Sunstein, The World vs. the United States and
China? The Complex Climate Change Incentives of the Leading Greenhouse Gas Emitters, 55 UCLA L. Rev. 1675
(2008); Michael P. Vandenbergh, Climate Change: The China Problem, 81 S. Cal. L. Rev. 905 (2008). We agree
with this consensus. See Eric A. Posner & David Weisbach, Climate Change Justice (2010); Daniel Abebe &
Jonathan S. Masur, International Agreements, Internal Heterogeneity, and Climate Change: The “Two Chinas”
Problem, 50 Va. J. Int’l L. 325 (2010).
3
See Perry Bacon, Jr., Lack of Votes for Senate Democrats’ Energy Bill May Mean the End, Wash. Post., July 23,
2010, available at http://www.washingtonpost.com/wp-dyn/content/article/2010/07/22/AR2010072203614.html
(“Conceding that they can't find enough votes for the legislation, Senate Democrats on Thursday abandoned efforts
to put together a comprehensive energy bill that would seek to curb greenhouse gas emissions . . . .”).
44
The EPA has taken the more limited step of requiring that major carbon dioxide emitters obtain permits. See
Prevention of Significant Deterioration and Title V Greenhouse Gas Tailoring Rule, 75 Fed. Reg. 31514 (Envtl.
Prot. Agency June 3, 2010). We discuss this regulation below. See infra note 71and accompanying text.
5
Massachusetts v. EPA, 549 U.S. 497 (2007).
The story begins in 2008, when the Ninth Circuit struck down the Department of
Transportation’s corporate average fuel economy (CAFE) standard for light trucks because the
cost-benefit analysis used by DOT to determine the standard failed to take account of its
beneficial effects for climate change.6 The petitioners, which included a number of
environmental groups, had tried to persuade the court that DOT should not have used costbenefit analysis, which they believed was impermissible under the statute. The court rejected
this argument but agreed with the petitioners that if DOT uses cost-benefit analysis, then it
cannot arbitrarily include some benefits (reduction in automobile noise and congestion, for
example) while excluding the benefits for the climate.7 On remand, DOT set to work monetizing
the climate benefits. Other agencies followed its path.
So it was cost-benefit analysis that initiated climate regulation at the federal level.
Environmentalists have frequently argued that cost-benefit analysis is a political tool that
government agencies use to avoid regulation.8 They have harshly criticized Executive Order
12,296, a Clinton-era executive order that requires regulatory agencies to issue cost-benefit
analyses for major regulations.9 Yet this order supplied DOT with the authority to engage in
cost-benefit analysis for the CAFE standards, which in turn led to judicial repudiation because
DOT’s cost-benefit analysis ignored climatic effects. In this instance, liberal defenders of costbenefit analysis, who have argued that it promotes regulation where regulation is warranted, have
been vindicated.10
Or have they? A close reading of the cost-benefit analyses performed by agencies in
connection with climate regulation reveals much to worry about. There is a wide gap between
the theory of cost-benefit analysis and the performance of the agencies.
In theory, the relevant monetary valuations for cost-benefit analysis of climate regulation
would come from climate science and from economics. The science establishes the approximate
effects of carbon emissions on the climate. The economics converts these effects into monetary
valuations based on their impact on people’s consumption patterns. Thus, one can establish that
an additional metric ton of carbon emissions will cause long-term disruptions to climate patterns
that will interfere with agriculture, raise sea levels, and so forth, which will either directly injure
people (for example, through an increase in the price of food) or require expensive remedial
measures (such as the construction of sea walls). Agencies should use this figure—known as the
social cost of carbon (“SCC”)—in cost-benefit analysis of regulations.
See Ctr. for Biological Diversity v. Nat’l Highway Traffic Safety Admin., 538 F.3d 1172, 1198-1204 (9th Cir.
2008).
7
Id. at 1201.
8
See, e.g., Frank Ackerman & Lisa Heinzerling, Priceless: On Knowing the Price of Everything and the Value of
Nothing (2004); Amy Sinden, Douglas A. Kysar & David M. Driesen, Cost-Benefit Analysis: New Foundations on
Shifting Sand, 3 Reg. & Governance 48, 50 (2009).
9
The Clinton order renews an earlier order issued by President Reagan, which may explain the association between
cost-benefit analysis and conservative, anti-regulatory sentiment. See Matthew Adler & Eric A. Posner, New
Foundations of Cost-Benefit Analysis 1-5 (2006).
10
See Richard L. Revesz & Michael A. Livermore, Retaking Rationality: How Cost-Benefit Analysis Can Better
Protect the Environment and Our Health (2008); Cass R. Sunstein, Risk and Reason: Safety, Law, and the
Environment (2002).
6
3
However, the science does not produce fine-grained predictions with a high level of
confidence. It is possible to predict that the median global temperature will rise but not that
temperatures or rainfall will be disrupted in a crucial agricultural region. Yet the latter
information is necessary for an adequate cost-benefit analysis. The economics poses even
greater difficulties. The three major models on which agencies rely are extraordinarily crude.
The cost of climate change will be high, but it is not clear how high, and one cannot conduct
cost-benefit analysis of a regulation without knowing what its economic effect will be.
The regulatory agencies recognize these problems and so far have performed curious
evasive maneuvers. They calculate a range of costs rather than a point estimate; then they
observe that their calculations depend on assumptions that are little more than guesses; and then
they announce that the specification of a precise figure is irrelevant because the regulation passes
a cost-benefit test given any of the possible social costs of carbon. (Agencies often refer to this
as a “sensitivity test”—they are testing whether the regulation is sensitive to the price placed on
alleviating carbon emissions.) For example, in a 2008 regulation of air conditioners and heat
pumps, the Department of Energy estimated that the social cost of carbon fell somewhere
between $0 and $20 per ton, and then declined to use any SCC figure in its cost-benefit analysis
on the ground that the proposed regulation passed a cost-benefit analysis regardless of the value
of the SCC.11 In a 2009 regulation of fuel economy standards, the Department of Transportation
calculated a separate $2 domestic SCC and a $33 global SCC, and then used these figures in a
sensitivity test but not in the cost-benefit analysis.12 Meanwhile, in 2008 the Environmental
Protection Agency estimated SCCs of $68 per ton (using a 2 percent discount rate) and $40 per
ton (using a 3 percent discount rate) in an advance notice of proposed rulemaking under the
Clean Air Act.13
In response to this confusion, the Office of Management and Budget convened an
Interagency Working Group (“IWG”) and gave it the task of resolving the disagreements among
agencies. The resulting document contains a highly sophisticated and carefully written survey of
the academic work, emphasizing that many uncertainties remain in the science and economics. It
proposes a range of SCCs reflecting different assumptions about the appropriate discount rate
and the year in which the regulation is issued.14 For 2010, the SCC is $4.70 at a discount rate of
5 percent, $21.40 at a discount rate of 3 percent, and $35.10 at a discount rate of 2.5 percent.
The IWG does not instruct agencies which discount rate to use. Since the IWG figures were
issued, agencies have stopped making their own calculations and have instead relied on the IWG
figures. But they have generally ignored the SCC when performing the cost-benefit analysis
used to justify their regulations and instead employed it only in sensitivity tests.
11
Commercial Standard Sized Packed Terminal Air Conditioners and Packed Terminal Heat Pumps, 73 Fed. Reg.
58,772 (Dep’t of Energy October 7, 2008).
12
Average Fuel Economy Standards Passenger Cars and Light Trucks Model Year 2011, 74 Fed. Reg. 14,196
(Dep’t of Transp., Nat’l Highway Traffic and Safety Admin. March 30, 2009).
13
Regulating Greenhouse Gas Emissions Under the Clean Air Act, 73 Fed. Reg. 44,354 (Envt’l. Prot. Agency
proposed July 30, 2008).
14
See Energy Conservation Program: Energy Conservation Standards for Small Electric Motors, 75 Fed. Reg.
10874-01 app. 15A at 40 (Dep’t of Energy Mar. 9, 2010) (Interagency Working Group on Social Cost of Carbon,
Social Cost of Carbon for Regulatory Impact Analysis Under Executive Order 12866) [hereinafter IWG Report].
4
This Article examines the record of the agencies (Part I) and the reasoning in the IWG
report (Part II). We write on a blank slate. Legal scholars have analyzed the prospects for an
international agreement on climate change;15 the possible forms of federal climate legislation;16
the future of EPA regulation under the Clean Air Act;17 and the possibility that greenhouse gases
might be regulated by states and localities,18 or through litigation.19 But the national regulatory
efforts already underway have escaped their notice.
Much of our Article is devoted to cataloging the errors in the analysis of the agencies and
the IWG report. But we also develop a larger theme about the relationship between cost-benefit
analysis and politics. Although we both support cost-benefit analysis for routine agency action,
we believe that some of the criticisms of cost-benefit analysis have plausibility in the context of
climate change. Climate change poses two problems for cost-benefit analysis. First, there is
tremendous uncertainty—not about whether climate change will harm human beings, but about
which human beings will be harmed, and where, and how much, and what weight those harms
should have in the cost-benefit calculation. In the absence of this information, agencies cannot
conduct proper cost-benefit analyses. Second, climate change is an international problem, while
E.g., Michael P. Vandenbergh & Mark A. Cohen, Climate Change Governance: Boundaries and Leakage, 18
N.Y.U. Envtl. L.J. 221 (2010) (describing mechanisms for bringing major polluting nations into compliance with
international climate accords); Michael P. Vandenbergh, Climate Change: The China Problem, 81 S. Cal. L. Rev.
905 (2008) (analyzing the international nature of global warming and the need for an international solution);
Jonathan B. Wiener, Think Globally, Act Globally: The Limits of Local Climate Policies, 155 U. Pa. L. Rev. 1961
(2007) (same); Cass R. Sunstein, Of Montreal and Kyoto: A Tale of Two Protocols, 38 Envtl. L. Rep. News &
Analysis 10566 (2008) (discussing the need for, and difficulties with, an international climate accord).
16
E.g., David A. Super, From The Greenhouse to the Poorhouse: Carbon-Emissions Control and the Rules of
Legislative Joinder, 158 U. Pa. L. Rev. 1093 (2010) (analyzing the leading legislative options for regulating carbon
emissions); Lisa Schultz Bressman, The Future of Agency Independence, 63 Vand. L. Rev. 599 (2010) (discussing
the ramifications of climate change legislation on agency structure and authority); Gilbert E. Metcalf & David
Weisbach, The Design of a Carbon Tax, 33 Harv. Envtl. L. Rev. 499 (2009) (describing how a legislatively enacted
federal carbon tax might be structured); Robert L. Glicksman, Balancing Mandate and Discretion in the Institutional
Design of Federal Climate Change Policy, 102 Nw. U. L. Rev. Colloquy 196 (2008) (analyzing the question of how
much authority Congress should delegate to the EPA to regulate climate change); Hari M. Osofsky, Climate Change
Legislation in Context, 102 Nw. U. L. Rev. Colloquy 245 (2008) (considering various legislative proposals).
17
E.g., J.D. Ruhl & James Salzman, Climate Change, Dead Zones, and Massive Problems in the Administrative
State: A Guide For Whittling Away, 98 Cal. L. Rev. 59 (2010) (noting the legal problems involved with agency
regulation of a massive issue such as climate change); Abigail R. Moncrieff, Reincarnating the “Major Questions”
Exception to Chevron Deference as a Doctrine of Noninterference (Or Why Massachusetts V. EPA Got It Wrong),
60 Admin. L. Rev. 593 (2008) (arguing that the Supreme Court should have refused to interfere with the EPA’s
decision not to consider regulating carbon dioxide); Madeline June Kass, A Nepa Climate Paradox: Taking
Greenhouse Gases Into Account in Threshold Significance Determinations, 42 Ind. L. Rev. 47 (arguing that federal
agencies should take climate change into account when drafting NEPA analyses) (2009); Jason Scott Johnston,
Climate Change Confusion and the Supreme Court: The Misguided Regulation Of Greenhouse Gas Emissions
Under The Clean Air Act, 84 Notre Dame L. Rev. 1 (2008) (arguing that the EPA should have been permitted to
regulate climate change).
18
E.g., Katherine A. Trisolini, All Hands On Deck: Local Governments and the Potential for Bidirectional Climate
Change Regulation, 62 Stan. L. Rev. 669 (2010) (suggesting the possibility of state or local regulation of greenhouse
gas production); Hari M. Osofsky, Is Climate Change “International”? Litigation's Diagonal Regulatory Role, 49
Va. J. Int'l L. 585 (2009) (discussing regulation at local, state, federal, and international levels).
19
E.g., Randall S. Abate, Public Nuisance Suits for the Climate Justice Movement: The Right Thing and the Right
Time, 85 Wash. L. Rev. 197 (2010) (proposing the use of private law as a means of forcing reductions in carbon
emissions); Kirsten H. Engel, Harmonizing Regulatory and Litigation Approaches To Climate Change Mitigation:
Incorporating Tradable Emissions Offsets Into Common Law Remedies, 55 U. Pa. L. Rev. 1563 (2007) (same).
15
5
American regulatory practice, including its use of cost-benefit analysis, is oriented to activities
with domestic effect. International relations pose inescapable political questions, which costbenefit analysis cannot resolve.
These shortcomings demonstrate the limits of cost-benefit analysis as a policy tool. The
best case for cost-benefit analysis is that its recommendations are politically neutral in the sense
of drawing on widely shared intuitions about human well-being.20 Cost-benefit analysis cannot
cope with inherently political questions involving contested normative issues. In some cases, the
source of conflict is so clear that no one seriously argues that regulatory agencies should solve
them using cost-benefit analysis. Consider abortion, affirmative action, and religious
accommodation. In other cases, the source of conflict is more subtle. We will argue that climate
change is one such case. In Part III, we make this argument and suggest ways forward.
I. Carbon Pricing in Existing Regulations
In the past two years, fourteen federal regulations have addressed the benefits of reduced
carbon dioxide emissions.21 Some of these regulations predate the IWG report and draw upon
other calculations of the social cost of carbon; others obtain their carbon prices directly from the
IWG report. Agencies that depend on pricing carbon run squarely into the problem that there is
no single price: the IWG, for instance, offers four different prices. The agencies attempt to skirt
the problem by claiming that their regulatory choices do not depend upon the precise cost of
carbon selected. They endeavor to prove that they would necessarily have opted for the identical
regulation under any of the possible carbon prices.
In the sections that follow, we survey three representative regulations, two promulgated
before the IWG report and one that employs the IWG’s carbon prices. These sections have two
purposes. First, we provide a detailed picture of how federal agencies have already taken up the
task of regulating carbon. Second, we demonstrate that for all of these regulations, the choice of
a carbon price is in fact significant. Agencies cannot simply pretend that their regulatory
decisions are invariant to the SCC; they must actually select a cost of carbon if they are to
regulate intelligently.
A. Model Year 2011 CAFE Standards
One of the earliest federal regulations to incorporate a social cost of carbon was the
Department of Transportation’s (“DOT”) new set of Corporate Average Fuel Economy
(“CAFE”) standards for Model Year 2011, issued on March 30, 2009.22 CAFE standards are
minimum requirements for the average fuel economy of a carmaker’s entire fleet, adjusted by the
number of each automobile sold. Thus, if Toyota sells 1000 pickup trucks, each of which
averages 20 miles/gallon, and 3000 compact cars, each of which averages 30 miles/gallon,
20
For a defense of this claim, see Adler & Posner, supra note 9.
We cite them, infra.
22
Average Fuel Economy Standards Passenger Cars and Light Trucks Model Year 2011, 74 Fed. Reg. at 14,196.
21
6
Toyota’s corporate average fuel economy is approximately 27 miles/gallon.23 If DOT issues a
CAFE standard higher than 27, then Toyota must adjust by withdrawing some of the pickup
trucks from the market, increasing its sales of compact cars, or improving the fuel efficiency of
the cars and/or trucks.
According to the DOT, the majority of the benefits from stricter CAFE standards stem
from savings in the production and distribution of gasoline: if drivers use less fuel, less money
need be spent on producing and distributing that fuel.24 Nonetheless, one of the most obvious
additional benefits of requiring that automakers produce fuel-efficient automobiles is the
reduction in greenhouse gas emissions. After the Ninth Circuit overturned DOT’s CAFE
standards for model years 2008-2011 for failing to monetize the benefit of reduced carbon
emissions, DOT went back to the drawing board.25 DOT solicited public comments and then
settled on a range of three prices for the SCC derived from a 2008 paper by Richard Tol, who, as
we will see, is one of the economists whose work the IWG incorporates.26 Tol’s 2008 paper is a
survey of over two hundred separate estimates of the social cost of carbon published in peerreviewed journals through 2006.27 DOT selected as its middle value the mean price from Tol’s
survey: $33 per ton of carbon dioxide.28 DOT’s “high” estimate is one standard deviation above
the mean from Tol’s study: $80 per ton of CO2.29
Like the IWG report, these prices include all global benefits, foreign and domestic. For
its “low” estimate, DOT selected its best estimate of the purely domestic benefits of reducing
carbon emissions.30 It calculated this price by adjusting Tol’s $33 estimate downward based on
the “U.S. share of world economic output (which ranges from 20-28 percent) and published
estimates of the relative sensitivity of the U.S. economy to climate changes.”31 DOT calculated
that the United States would suffer between 0 and 14 percent of the global costs of climate
change, chose the midpoint of that range (7 percent), and calculated the domestic cost of carbon
as 7% of $33, or approximately $2. That left DOT with three values—$2, $33, and $80—that
are roughly similar to the carbon prices subsequently calculated by the IWG.32
23
Corporate average fuel economy is calculated using the harmonic mean of the fuel economies of the fleet vehicles,
rather than the arithmetic mean (which would yield a corporate average fuel economy of 27.5 miles/gallon using the
hypothesized
numbers).
See
Corporate
Average
Fuel
Economy
Wikipedia,
http://en.wikipedia.org/wiki/Corporate_Average_Fuel_Economy#Calculuation (last visited July 8, 2010).
24
74 Fed. Reg. at 14,412 (Table IX-3). According to the DOT, these savings comprise approximately three quarters
of the total benefit of fuel economy regulation.
25
See Ctr. for Biological Diversity v. Nat’l Highway Traffic Safety Admin., 538 F.3d 1172 (9th Cir. 2008).
26
Richard S.J. Tol, The Social Cost of Carbon: Trends, Outliers, and Catastrophes, Economics—the Open-Access,
Open-Assessment E-Journal, 1, 2 (2008).
27
Id.
28
Average Fuel Economy Standards Passenger Cars and Light Trucks Model Year 2011, 74 Fed. Reg. at 14,346.
29
Id. It is worth noting that one standard deviation from the mean is approximately the 85th percentile of the
distribution. Accordingly, this $80 per ton price represents only an 85th percentile estimate, not an upper bound.
30
Id.
31
Id. at 14,349; see also Mark A. Delucchi, Summary of the Non-Monetary Externalities of Motor Vehicle Use,
UCD-ITS-RR-96-3 (9) rev. 1, Institute of Transportation Studies, University of California, Davis, originally
published
September
1998,
revised
October
2004,
pp.
49-51.
Available
at
http://
www.its.ucdavis.edu/publications/2004/UCD-ITS-RR-96-03(09)_rev1.pdf (last accessed March 23, 2009).
32
See infra Part II.
7
These initial carbon prices were denoted in 2007 dollars. DOT decided to increase the
prices by 2.4% in each subsequent year “because the increased pace and degree of climate
change—and thus the resulting economic damages—caused by additional emissions are both
expected to rise in proportion to the existing concentration of CO2 in the earth’s atmosphere.” 33
Crucially, however, these 2.4% increases were not pegged to constant dollars. In other words, if
the value of one ton of carbon dioxide were $1.00 in 2007 (in 2007 dollars), it would be $1.024
in 2008 (in 2008 dollars). In order to determine the value of that 2008 ton of carbon dioxide in
constant (2007) dollars, it would be necessary to discount the $1.024 price back to 2007 dollars.
DOT elected to apply a 3% discount rate to the social costs of carbon.34 This meant that in
constant dollars the social costs of carbon would decrease over time. This approach contrasts
with the IWG report, which has carbon prices that increase steadily over time (in constant
dollars), a point to which we return below.35
The DOT then ran a set of models meant to estimate the costs and benefits of different
CAFE standards and selected what it believed to be the “optimized” rule—the one that would
generate the greatest net benefits. The agency tested this “optimized” rule to determine whether
it was sensitive to the choice of carbon price, and concluded that “the optimized CAFE standards
for MY 2011 cars and light trucks were unaffected by the choice among those values for
reducing CO2 emissions from fuel production and use.”36 DOT reported the results of its costbenefit analysis for this optimized rule, along with rules 25% below (i.e., lower mandatory fuel
efficiency) and 25% and 50% above (i.e., higher mandatory fuel efficiency) the optimal. DOT
also reported cost-benefit numbers for two other potential standards: 1) the standard that would
generate costs approximately equal to benefits; and 2) the standard at the point of “technical
exhaustion”—the greatest fuel efficiency possible given current technology. The results of
DOT’s cost-benefit analysis are reproduced in Table 1 below:
Table 1: Benefits and Costs of Possible CAFE Standards (versus Baseline of No Action) Over Lifetime of Model
Year 2011 Automobiles (millions of 2007 dollars)37
25% Below
Optimized 25% Above 50% Above Total Costs Technology
Optimized
Optimized
Optimized
= Total
Exhaust
Benefits38
Total
1,707
1,948
2,321
2,763
3,676
9,356
Benefits
Total
940
1,145
1,918
2,545
4,009
18,120
Costs
Net
767
802
403
218
-334
-8,765
Benefits
As part of this cost-benefit analysis, DOT estimated that its optimized regulation would
eliminate 8.33 million tons of CO2 emissions over the lifetime of the model year 2011
33
Average Fuel Economy Standards Passenger Cars and Light Trucks Model Year 2011, 74 Fed. Reg. at 14,350.
Id. at 14,355.
35
For instance, the IWG’s median carbon prices (discounted at 3%) grow at an approximate yearly rate of 2.1%, in
constant dollars, between 2010 and 2020. IWG Report at 40.
36
Average Fuel Economy Standards Passenger Cars and Light Trucks Model Year 2011, 74 Fed. Reg. at 14,351.
37
Id. at 14,385 (Tables VII-4 through VII-6).
38
As is obvious from the numbers, costs are not quite equivalent to benefits under this regulatory option.
34
8
automobiles.39 It priced these emissions at $2/ton, for a total benefit of approximately $16
million. It is immediately apparent that if the social cost of carbon were much higher—$80 per
ton, or even $64.9 per ton, the 95th percentile value from the SCC—the benefits from reduced
CO2 emissions would exceed the gap between the Optimized and 25% Above Optimized or 50%
Above Optimized standards. Of course, in order to determine whether one of those options
would in fact produce greater net benefits than the Optimized standard, it is necessary to know
what reduction in emissions each of these alternatives would produce. To the best of our
knowledge, DOT did not report these figures for the lifetime of model year 2011 automobiles.
However, it did report expected reductions in CO2 emissions from 2010 to 2100, assuming that
its regulatory standards remained in force during that time. If we assume that emissions
reductions from 2010 to 2100 are proportional to emissions reductions for model year 2011
automobiles, we can calculate expected reductions for model year 2011 vehicles under each
regulatory scenario. Those results and calculations are listed in Table 2 below:
Table 2: CO2 Emissions by Passenger Cars and Light Trucks with Alternative CAFE Standards for Model Year
2011 (million tons CO2)40
Emissions Reductions
Emissions Reductions from
Over the Lifetime of
Cumulative Emissions,
Alternative
2010-2100 Compared to No
Model Year 2011
2010-2100
Action Alternative
Automobiles
No Action
210,279
0
0
25% Below Optimized
209,076
1,203
7.44*
Optimized
208,932
1,347
8.33
25% Above Optimized
208,743
1,536
9.50*
50% Above Optimized
208,440
1,839
11.37*
Total Costs = Total Benefits
208,015
2,265
14.01*
Technology Exhaustion
204,228
6,052
37.43*
Consider the results of this analysis if the social cost of carbon is priced at the IWG’s 3%
discount value ($21.40 per ton in 2010). Let us suppose that the lifetime emissions for model
year 2011 automobiles are evenly distributed across the years 2010 through 2021. If that is the
case, the costs and benefits are as follows:41
Id. at 14,412.
Id. at 14,390 (Table VII-11). Values labeled with “*” were calculated, not reported from the DOT regulation.
41
DOT also claimed that reducing American oil consumption would produce “monopsony benefits”—lower oil
prices for consumers due to the fact that a reduction in American monopsony consumption of oil would decrease
world oil prices. See id. at 14,329. However, DOT decided to significantly reduce its estimation of these
monopsony benefits when pricing CO2 emissions at $33 and $80 per ton. See id. at 14,416 n. 473. DOT did not
explain its reason for doing so, and we cannot make sense of this decision.
39
40
9
Table 3: Benefits and Costs of Possible CAFE Standards (versus Baseline of No Action) Over Lifetime of Model
Year 2011 Automobiles (millions of 2007 dollars) at IWG 3% Discount Carbon Prices
25% Below Optimized 25% Above 50% Above Total Costs Technology
Optimized
Optimized
Optimized
= Total
Exhaust
Benefits
Total
1,868
2,128
2,526
3,008
3,978
1,0164
Benefits
Total Costs
940
1,145
1,918
2,545
4,009
18,120
Net
948
983
606
463
-31
-8,044
Benefits
The higher price on carbon does not alter the benefits so radically as to make regulations
25% Above or Below Optimized preferable. But the gap in net benefits between these regulatory
choices has narrowed. Moreover, the “Optimized” CAFE standard is meant to be the regulation
that provides the greatest net benefits. With a different price on carbon, that optimal point may
well shift upward—perhaps not all the way to 25% Above the current point, but at least
somewhat above DOT’s chosen standard.
In addition, these costs and benefits reflect model year 2011 cars only. If the regulations
remained in place and costs and benefits were computed for the next 10, 20, or 50 years, they
would generate even greater reductions in carbon emissions. If those reductions were valued in
accordance with the IWG report, the “optimal” CAFE standards would be more stringent (that is,
require greater fuel efficiency) than DOT’s calculations reveal. This is due in large part to the
fact that under the IWG report the discounted social costs of carbon increase over time, while
DOT proceeded on the assumption that the discounted social costs will diminish with time.42
Thus, if DOT’s claim that its choice of CAFE standard was invariant to the price of carbon is
correct, it is likely correct only under a particular set of parameters specified by DOT itself. As
the price of carbon and the regulatory time frame increase, stricter CAFE standards appear more
attractive.
B. Energy Conservation Standards for Fluorescent and Incandescent Lamps
On July 14, 2009, the Department of Energy (“DOE”) promulgated a regulation setting
energy efficiency standards for General Service Fluorescent Lamps (“GSFL”) and Incandescent
Reflector Lamps (“IRL”).43 More efficient fluorescent and incandescent lamps reduce electricity
consumption, and thus simultaneously eliminate greenhouse gas emissions.44 In fact, according
to DOE’s analysis, the reduction in carbon emissions is one of the primary sources of benefits
from mandating more efficient lamps.45 However, because this regulation was completed before
42
IWG Report at 40. In addition, the lower the discount rate, the faster the rate of increase in the price of carbon.
Thus, the lower the SCC discount rate, the greater the gap over time between benefits calculated according to the
IWG report and benefits calculated by DOT.
43
Energy Conservation Program: Energy Conservation Standards and Test Procedures for General Service
Fluorescent Lamps and Incandescent Reflector Lamps, 74 Fed. Reg. 34,080 (Dep’t of Energy July 14, 2009).
44
The principal cost of such regulation is the greater expense required to produce more efficient lamps, a cost that
will most likely be passed along to consumers. See id. at 34,084 (predicting higher consumer prices).
45
Reducing consumption of electricity also produces other environmental benefits, such as reducing emissions of
sulfur dioxide, mercury, and nitrous oxide. Id. We focus here on carbon dioxide, as it is responsible for the lion’s
share of the benefits and is most subject to uncertainty in pricing.
10
the publication of the IWG report, DOE was also forced to rely upon other sources when pricing
the benefits of lower emissions. It chose to employ the Department of Transportation’s prices
from its CAFE standards: $2, $33, and $80 per ton.46 DOE also decided to increase these values
by 3% per year “to represent the expected increases, over time, of the benefits associated with
reducing CO2 and other greenhouse gas emissions.”47 As with the CAFE standards, however,
these 3% increases were not pegged to constant dollars. DOE applied two discount rates to the
benefits and costs of its regulation: 3% and 7%. Thus, at 3%, the social costs of carbon are
constant (in constant dollars) over time; at 7% they decrease fairly rapidly. This is again in
contrast to the IWG, which concluded that the social cost of carbon will increase steadily over
time.
For each of these two types of lamps, DOE considered five different potential levels of
energy efficiency, numbered in order of increasing stringency. That is, the higher the number,
the greater the energy efficiency that would be required. DOE calculated the benefits of reduced
carbon emissions for the two lamps at each of these five potential levels of efficiency, and at the
3% and 7% discount rates. DOE’s estimates of reduced CO2 emissions are presented as ranges,
due to DOE’s inability to predict whether more efficient lamps will obviate coal-fired or natural
gas-fired plants.48 In the interests of simplicity we focus here only on IRL, though the analysis is
the same for both types of lamps. DOE’s results are displayed below:
Table 4: Estimates of Value of CO2 Emissions Reductions for IRL by Level of Regulation at SevenPercent and Three-Percent Discount Rates49
Regulation
Estimated
Value of estimated CO2 emission
Value of estimated CO2 emission
Level
cumulative CO2 reductions (billion 2008$) at 7%
reductions (billion 2008$) at 3%
discount rate
discount rate
emission
reductions from
2012 to 2042 CO value CO value CO value CO value CO value CO value
2
2
2
2
2
2
(MMt)
of $2/ton of $33/ton of $80/ton of $2/ton of $33/ton of $80/ton
1
7 to 20
0.0 to 0.0
0.1 to 0.3
0.3 to 0.8
0.0 to 0.0 0.3 to 0.7
0.6 to 1.7
2
19 to 49
0.0 to 0.1
0.4 to 0.8
0.8 to 2.1
0.0 to 0.1 0.7 to 1.7
1.6 to 4.1
3
38 to 85
0.0 to 0.1
0.7 to 1.5
1.7 to 3.6
0.1 to 0.2 1.3 to 2.9
3.2 to 7.1
4
44 to 106
0.0 to 0.1
0.8 to 1.8
1.9 to 4.4
0.1 to 0.2 1.5 to 3.7
3.7 to 8.9
5
53 to 118
0.1 to 0.1
1.0 to 2.0
2.3 to 4.9
0.1 to 0.2 1.8 to 4.1
4.5 to 9.9
DOE then calculated the annual benefits (net of costs) at each level of regulation. It
offered “primary,” “low,” and “high” estimates of these benefits. The “primary” estimate is
DOE’s best guess, given an intermediate price of carbon (approximately $33/ton) and median
estimates of the reduction in emissions. The “low” and “high” estimates are based on lower and
higher estimates of both price and emissions. DOE reported net benefits at discount rates of 3%
See supra Part II.A.1.
Energy Conservation Program: Energy Conservation Standards and Test Procedures for General Service
Fluorescent Lamps and Incandescent Reflector Lamps, 74 Fed. Reg. at 34,163.
48
Coal-fired plants produce greater CO2 emissions per unit of electricity.
49
74 Fed. Reg. 34,080, 34,164 (Table VII.28).
46
47
11
and 7%. The results are displayed below, with the level of regulation that would provide the
highest net benefit (for each emissions estimate and discount rate) in bold:
Table 5: Annualized Benefits Net of Costs for IRL by Level of Regulation (in $millions/year, for the period 20122042)50
Regulation
Primary estimate at
Low estimate at discount High estimate at discount
Level
discount rate:
rate:
rate:
7%
3%
7%
3%
7%
3%
1
18
29
-9
-2
44
61
2
326
352
203
215
449
489
3
459
532
297
339
620
725
4
532
590
179
207
885
973
5
621
687
247
280
994
1093
These tables make clear that the optimal level of regulation depends upon the price of
carbon and estimated emissions reductions. Consider Table 5, which reports the net benefits of
IRL regulation. At low estimates of emissions and price, Level 3 regulation provides the greatest
net benefits. Indeed, the advantages over Levels 4 and 5 are quite pronounced. But at primary
or high estimates, Level 5 regulation is optimal and provides substantially greater net benefits
than Levels 3 or 4. In addition, it is hard to know where the crossover point lies. There are
obviously carbon prices and emissions estimates above the “low” estimates at which Level 3 is
optimal; and there are also prices and emissions estimates below the “primary” estimates at
which Level 5 is superior. Indeed, there may be some combination of price and emissions
between the low and primary estimates at which Level 4 regulation provides the greatest net
benefits.
In the end, DOE opted to regulate IRL (and GSFL) at Level 4, despite the cost-benefit
advantages of requiring greater energy efficiency.51 The agency concluded that Level 5
regulation was “economically [un]justified,” contrary to the results of its own cost-benefit
analysis, because it would place excessive burdens on producers and some consumers who
would be forced to pay higher prices for more efficient lamps.52 In some sense, then, the choice
of a price for carbon was irrelevant to the final regulatory decision. But if the agency had
followed its own cost-benefit analysis, the particular price put on carbon emissions would have
been important—and perhaps decisive.
C. Energy Conservation Standards for Small Electric Motors
The first regulation to employ the IWG’s SCC was the Department of Energy’s Energy
Conservation Standards for Small Electric Motors, promulgated on March 9, 2010.53 The IWG
Id. at 34,171 (Table VII.34).
Id. at 34,169-73.
52
Id. (citing 42 U.S.C. § 6295(o)(2)(A)). We have our doubts about the normative soundness of this conclusion.
See Jonathan S. Masur & Eric A. Posner, Against Feasibility Analysis, 77 U. Chi. L. Rev. 657 (2010).
53
Energy Conservation Program: Energy Conservation Standards for Small Electric Motors, 75 Fed. Reg. 10,874
(Dep’t of Energy March 9, 2010).
50
51
12
reportwas first published as an appendix to this rule,54 and to our knowledge it has not been
published again since. As the title would indicate, this regulation mandates energy efficiency
guidelines for two types of small electric motors: “polyphase” small electric motors and
“capacitor-start” small electric motors.55 For each of these two types of motors, DOE considered
eight different potential levels of energy efficiency, numbered in order of increasing stringency.
(Again, the higher the number, the greater the energy efficiency that would be required.) The
regulatory levels for capacitor-start motors are numbered 1 through 8; the levels for polyphase
motors are numbered 1 through 7, with an additional level “4b” that represents an adjusted set of
regulations that the agency believed would more closely approach optimal levels. DOE then
examined the effect of each potential regulation on carbon dioxide emissions from 2015, when
the regulation will take effect, until 2045.56 It estimated the likely reductions in CO2 emissions
that would result from increased efficiency (and thus diminished need for electrical power
generation), and then priced the benefits of these reduced emissions using the four different
carbon prices reported by the IWG. Again, in the interests of simplicity we focus here only on
polyphase motors, though our analysis of capacitor-start motors is essentially the same. DOE’s
results are as follows:
Table 6: Estimates of Global Present Value of CO2 Emissions Reductions for the Period 2015-2045
Under Polyphase Small Electric Motor Trial Standard Levels57
Regulation
Level
Estimated
cumulative CO2
emission
reductions, Mt
Global value of CO2 emission reductions, million$ 2009
5% discount rate
3% discount rate 2.5% discount rate
1
2.3
8
2
4.6
3
8.3
4
3% discount rate,
95th percentile
40
68
122
16
81
138
248
28
146
248
445
9.3
32
165
280
502
4b
15.4
52
272
462
828
5
18.3
62
323
550
986
6
19.5
66
344
585
1049
7
21.2
72
375
638
1144
As an initial matter, it is a bit difficult to determine whether DOE has in fact priced its
own carbon reductions correctly. Take, for instance, the very first row of Table 6 (Level 1
regulation). DOE estimated that this level of regulation would reduce CO2 emissions by 2.3
million tons from 2015 to 2045. According to the IWG report, the price of one ton of CO2 in
2015 is $5.70 at a 5% discount rate.58 That price only rises over the next 30 years, reaching
$14.20 in 2045. In addition, the IWG prices are in 2007 dollars, while the dollar values reported
See id. at 10,911 (referencing “Appendix A of the Annex to Chapter 15 of the Technical Support Document”).
Id. at 10,932. For our purposes, the differences between these two types of motors are irrelevant. We mention
them both only to explain that the agency’s analysis proceeded in two parts.
56
Id. at 10,931.
57
Id. at 10,931 (Table VI.31).
58
IWG Report at 40.
54
55
13
in the DOE regulation are in 2009 dollars, which are less valuable. Thus, the global value of
CO2 emission reductions at Level 1 should be at minimum $13 million, and actually much
more.59 Yet the regulation values these CO2 reductions at only $8 million. This problem exists
with every one of the calculations, and we cannot explain these discrepancies.
DOE then calculated the benefits of regulation (net of costs) for the two types of motors,
at eight possible levels of regulation, across the four standard prices for carbon dioxide from the
IWG report. The results are displayed in Table 7 (polyphase motors) . The level of regulation
that will provide the highest net benefit at each price of carbon is in bold.
Table 7: Total Net Benefits at Various Social Costs of Carbon for Polyphase Small Electric Motors (2015-2045)60
Regulation Level
Consumer NPV (in billion$ 2009) at 3% discount rate with:
CO2 value of
$4.7/metric ton CO2
CO2 value of
$21.4/metric ton CO2
CO2 value of
$35.1/metric ton CO2
CO2 value of
$64.9/metric ton CO2
1
0.27
0.30
0.33
0.39
2
0.57
0.64
0.69
0.81
3
1.04
1.16
1.27
1.47
4
1.08
1.22
1.34
1.57
4b
1.49
1.73
1.92
2.29
5
0.83
1.11
1.34
1.79
6
0.13
0.42
0.66
1.14
7
(12.57)
(12.26)
(11.99)
(11.47)
As the reader can observe, the level of regulation that provides the greatest net benefits is
not affected by the choice of carbon price. For polyphase motors, Level 4b delivers the greatest
net benefits irrespective of the price of carbon dioxide. It is on the basis of this finding that DOE
concluded that the choice of regulatory level was invariant to the choice of carbon price.
Yet it is clear that this conclusion is an artifact of the particular potential levels of
regulation DOE chose to analyze. If DOE had not included Level 4b in the analysis, its choice of
regulation would depend on the price it attached to CO2 emissions. Per Table 7, if CO2
emissions are priced at $4.70 or $21.40 per metric ton, Level 4 provides the greatest net benefits.
If emissions are priced at $35.1 per ton, Levels 4 and 5 provide equivalent benefits. And if
emissions are priced at $64.90 per ton, Level 5, a more energy-efficient standard, provides the
greatest benefits. DOE evades a decision between Levels 4 and 5 by settling on an intermediate
option—Level 4b—that is superior to both. But this raises the implication that there may be
other intermediate possibilities—imagine Level “4a” or “4c”—that are superior to Level 4b at
some prices of carbon and inferior at others.
59
2.3 million tons of CO2 at $5.70 comes to $13.11 million. However, the overall figure must be higher because 1)
the regulation reports values in 2009 dollars while the IWG report lists carbon dioxide prices in 2007 dollars; and 2)
the social costs of carbon are greater than $5.70 in every year subsequent to 2015.
60
75 Fed. Reg. 10,874, 10,934 (Table VI.41).
14
Once again, it is the agency’s decision to consider only particular discrete regulatory
options, rather than an entire continuum of possibilities, that allows it to conclude that it should
impose the same regulation regardless of the price of carbon. The agency chooses its options—
for example, Level 4b, but not Levels “4a” or “4c”—such that one level of regulation is superior
to all others at each of the four carbon prices. But that is quite different than saying that the
optimal level of regulation is unrelated to the price of carbon. In fact, the agency was most likely
compelled to investigate Level 4b precisely because the choice between Level 4 and Level 5
depended entirely on the social cost of carbon. The Department of Energy, like the Department
of Transportation, will not be able to avoid choosing a price for carbon dioxide emissions when
making regulatory decisions.
D. Further Agency Regulation Involving the SCC
We have discussed three of the 14 regulations that report an SCC. Of the remaining
regulations, seven involved the Department of Energy (air conditioners and heat pumps;
commercial refrigerators and freezers; residential gas ranges and ovens; commercial heating, air
conditioning, and water-heating equipment; refrigerated bottled or canned beverage vending
machines; residential water heaters, direct heating equipment, and pool heaters; and commercial
clothes washers61). Two of the remaining regulations involved the Department of Transportation
(applications for funding under the American Recovery and Reinvestment Act, and for funding
for national infrastructure).62 One involved EPA (fuels and fuel additives).63 And one involved
both DOT and EPA (vehicle emissions).64
All of the regulatory impact statements accompanying these regulations resemble the
three we have discussed. They typically report a range of SCCs, with the statements for the most
61
Energy Conservation Program for Commercial and Industrial Equipment: Packaged Terminal Air Conditioner
and Packaged Terminal Heat Pump Energy Conservation Standards, 73 Fed. Reg. 58,772-01 (Dep’t of Energy Oct.
7, 2008); Energy Conservation Standards for Commercial Ice-Cream Freezers; Self-Contained Commercial
Refrigerators, Commercial Freezers, and Commercial Refrigerator-Freezers Without Doors, 74 Fed. Reg. 1,092
(Dep’t of Energy Jan. 9, 2009); Energy Conservation Standards for Certain Consumer Products and for Certain
Commercial and Industrial Equipment, 74 Fed. Reg 16,040 (Dep’t of Energy Apr. 8, 2009); Energy Conservation
Standards and Test Procedures for Commercial Heating, Air-Conditioning, and Water-Heating Equipment, 74 Fed.
Reg. 36,312-01 (Dep’t of Energy July 22, 2009); Energy Conservation Standards for Refrigerated Bottled or Canned
Beverage Vending Machines, 74 Fed. Reg. 44,914-01 (Dep’t of Energy Aug. 31, 2009); Energy Conservation
Standards for Residential Water Heaters, Direct Heating Equipment, and Pool Heaters, 75 Fed. Reg. 20,112-01
(Apr. 16, 2010); Energy Conservation Standards for Certain Consumer Products and for Certain Commercial and
Industrial Equipment, 75 Fed. Reg. 1122-01 (Jan. 8, 2010).
62
Notice of Funding Availability for Supplemental Discretionary Grants for Capital Investments in Surface
Transportation Infrastructure Under the American Recovery and Reinvestment Act, 74 Fed. Reg. 28,755-01 (Dep’t
of Transp. June 17, 2009); Notice of Funding Availability for the Department of Transportation's National
Infrastructure Investments Under the Transportation, Housing and Urban Development, and Related Agencies
Appropriations Act for 2010, 75 Fed. Reg. 30,460-02 (Dep’t of Transp. June 1, 2010).
63
Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program, 75 Fed. Reg. 14,670-01
(Envtl. Prot. Agency Mar. 26, 2010).
64
Light-Duty Vehicle Greenhouse Gas Emission Standards and Corporate Average Fuel Economy Standards, 75
Fed. Reg. 25,324 (Envtl. Prot. Agency and Dep’t of Transp. May 7, 2010).
15
recent regulations using the IWG’s figures. And they all exclude the SCCs from the actual costbenefit analysis, instead merely reporting them or using them in a sensitivity analysis.65
Meanwhile, the EPA has begun deliberating about climate regulation under the Clean Air
Act. In Massachusetts v. EPA,66 the Supreme Court held that if anthropogenic carbon dioxide is
leading to global warming and causing harm to humans, it is an “air pollutant” under the Clean
Air Act and the EPA must regulate it.67 On December 7, 2009, the EPA announced its
conclusion that greenhouse gases threaten public health.68 This finding authorizes the EPA to
regulate carbon dioxide emissions in a variety of ways, including requiring that CO2 emitters to
install the “best available” technology to mitigate emissions. Such regulation would impose
significant costs on the American economy, but it might also produce substantial benefits by
mitigating the negative effects of global warming. Its effects would overshadow the comparably
trivial regulations that we have discussed so far.
The EPA is at a very early stage in the process of regulating carbon dioxide under the
Clean Air Act—it has not published proposed regulatory text or officially solicited comments
from interested private parties.69 In fact, the Obama administration has indicated that the EPA’s
finding is principally intended to compel Congress to pass greenhouse gas legislation.70 The
EPA has taken the more limited step of requiring that major CO2 emitters—those that produce
more than 100,000 tons of carbon dioxide per year—obtain permits and install the best available
technology before initiating new sources of pollution.71 However, in the cost-benefit analysis
that accompanied that regulation the EPA refused to quantify the benefits of reduced carbon
emissions, deeming them too uncertain.72 It did not so much as mention the SCC. If Congress
does not act and the EPA eventually promulgates much broader carbon dioxide regulation, it will
be essential that the agency conduct a full cost-benefit analysis in order to ascertain whether such
regulation is justified. An essential component of that cost-benefit analysis, perhaps the
preeminent part, will be a calculation of the benefits of reduced carbon emissions.
65
In some cases, the regulations ask for applications for funds, and ask applicants to use an SCC in determining the
environmental effects of their proposed project.
66
Massachusetts v. E.P.A., 549 U.S. 497 (2007).
67
Id. at 533.
68
News Release, Envtl. Prot. Agency, EPA: Greenhouse Gases Threaten Public Health and the Environment /
Science overwhelmingly shows greenhouse gas concentrations at unprecedented levels due to human activity (Dec.
7, 2009), available at http://yosemite.epa.gov/opa/admpress.nsf/7ebdf4d0b217978b852573590040443a/
08d11a451131bca585257685005bf252!OpenDocument.
69
See 5 U.S.C. § 553 (describing the rulemaking process that the EPA must undertake before regulating).
70
See Sindya N. Bhanoo, The E.P.A. Announces a New Rule on Polluters, N.Y. Times, May 14, 2010, at A13,
available at http://www.nytimes.com/2010/05/14/science/earth/14permit.html.
71
Prevention of Significant Deterioration and Title V Greenhouse Gas Tailoring Rule, 75 Fed. Reg. 31514 (Envtl.
Prot. Agency June 3, 2010). We describe this regulation as more limited both because it regulates only a small
subset of carbon emitters and because the impact of requiring that even a large emitter install the “best available”
technology will likely be “relatively small.” Id. at 31600. This is “due to the lack of available capture and control
technologies” for carbon dioxide. Id.
72
Id. at 31598-31601.
16
II. Problems
In this Part, we turn our attention to the Interagency Working Group’s report on the
Social Cost of Carbon, the definitive federal administrative statement on the subject. We
demonstrate that it suffers from a variety of problems of various types, problems that render its
conclusions unconvincing. At a fundamental level, the IWG’s error was in failing to recognize
the political nature of certain issues, treating them instead as technical matters; and in failing to
recognize the technical nature of other issues, seeking political solutions where none were
available.
A.
The IWG Report
We begin with a discussion of the IWG’s method for pricing carbon. The IWG’s
estimate of the social costs of carbon is based on predictions derived from three separate
computer models of climate change and economic harm. These models, all developed by
academics and widely used in estimating future climate harms, are known as DICE (“Dynamic
Integrated Climate and Economy”),73 PAGE (“Policy Analysis of the Greenhouse Effect”),74 and
FUND (“Climate Framework for Uncertainty, Negotiation, and Distribution”).75 The three
models differ in some significant respects (several of which we describe below), but for the most
part they operate in a similar fashion. The user enters a set of economic parameters, including
pre-existing baseline projections of economic growth and technological improvements,
developed within the standard economic literature.76 These projections include predictions of
future greenhouse gas emissions, which are a function of GDP and a society’s “carbon
intensity”: the amount of carbon a nation’s economy must generate in order to produce wealth.77
From these projected emissions, the climate models predict changes in the concentration
of greenhouse gases in the atmosphere.78 Those changes in greenhouse gas concentrations are, in
turn, used to predict changes in temperature, and the models then project economic harms (in the
form of diminished worldwide GDP) from the expected temperature increases.79 These models
thus involve both climate science—translations of greenhouse gas concentrations into
temperature changes—and environmental economics—predictions of the effect of warming on
73
DICE was developed by William Nordhaus, an economist at Yale University. William Nordhaus, A Question of
Balance: Weighing the Options on Global Warming Policies (2008); William Nordhaus and John Boyer, Warming
the World: Economic Models of Global Warming (2000).
74
PAGE was developed by Chris Hope, of the Cambridge University Judge Business School. Chris Hope, Optimal
Carbon Emissions and the Social Cost of Carbon Under Uncertainty, 8 The Integrated Assessment Journal 107
(2008); Chris Hope, The Marginal Impact of CO2 From PAGE2002: An Integrated assessment Model Incorporating
the IPCC’s Five Reasons for Concern, 6 The Integrated Assessment Journal 19 (2006).
75
FUND was developed by Richard Tol, a professor at the Economic and Social Research Institute in Dublin.
Richard Tol, Estimates of the Damage Costs of Climate Change. Part I: Benchmark Estimates, 21 Environmental
and Resource Economics 47 (2002); Richard Tol, Estimates of the Damage Costs of Climate Change. Part II:
Dynamic Estimates, 21 Environmental and Resource Economics 135 (2002).
76
The IWG report employed five sets of parameters derived from the Stanford Energy Modeling Forum exercise.
See IWG Report at 16.
77
Abebe & Masur, supra note. Not surprisingly, carbon intensity is primarily a function of a society’s level of
technology: higher-technology societies are able to produce greater wealth with lower pollution. Id.
78
IWG Report at 6.
79
Id.
17
GDP. In order to generate values for the social costs of carbon, the Interagency Group ran the
three models using standard baseline projections of economic growth and technological
development (the parameters described above) and determined the models’ predicted effects of
warming on GDP. The models are all probabilistic, and thus they each generate a probability
distribution of possible outcomes. The Interagency Group obtained the mean outcome for each
model, and then averaged these three means. The Group then re-ran the same models using the
same baseline projections, but with one additional ton of carbon emissions, in order to determine
the marginal effect on global GDP of that additional unit of carbon. The average reduction in
GDP across all three models from this second run, minus the average reduction in GDP from the
first (baseline) run, is the social cost of carbon: the amount of money saved for every marginal
ton of atmospheric carbon that is not emitted.
Rather than declare a single value for the social cost of carbon, the IWG reports different
values for every year from 2010 to 2050. The yearly values differ both because the cost of
carbon must be discounted to present value, and because one additional unit of carbon will likely
affect the environment differently in 2020 than it does 2010. This is due to the fact that the
relationship between atmospheric concentrations and carbon is not linear, and there will likely be
more carbon in the atmosphere in 2020 than in 2010. Accordingly, an additional ton of carbon
will likely cause greater warming in 2020 than it would in 2010 because it will combine with an
already carbon-saturated atmosphere.
The IWG does not specify a discount rate.80 Instead, the IWG reports average social
costs of carbon at discount rates of 5%, 3%, and 2.5%. The IWG also reports the 95th percentile
probabilistic value for the social cost of carbon (at a 3% discount rate), as a means of providing
something of an upper bound on the likely costs. The IWG predicts that the additional harm
from carbon emissions in later years will outpace the rate of monetary discounting, even at a
discount rate of 7%, and thus the social costs of carbon increase with time in constant dollars,
rather than decreasing. Table 8 reproduces the Interagency Group’s chart of social costs of
carbon at various discount rates from 2010 through 2021.
80
IWG Report at 17-23.
18
Table 8: Annual SCC Values: 2010–2050 (in 2007 dollars per ton)
Discount rate:
5%
3%
2.5%
95th
Year
2010
4.7
21.4
35.1
64.9
2011
4.9
21.9
35.7
66.5
2012
5.1
22.4
36.4
68.1
2013
5.3
22.8
37.0
69.6
2014
5.5
23.3
37.7
71.2
2015
5.7
23.8
38.4
72.8
2016
5.9
24.3
39.0
74.4
2017
6.1
24.8
39.7
76.0
2018
6.3
25.3
40.4
77.5
2019
6.5
25.8
41.0
79.1
2020
6.8
26.3
41.7
80.7
2021
7.1
27.0
42.5
82.6
The use of these carbon prices in federal regulations is meant to be straightforward.
Suppose a regulation will reduce carbon emissions by 10,000 tons per year in every year from
2010 through 2020. An agency would simply multiply the emissions avoided by the price of a
ton of emissions in the appropriate year to calculate the full carbon-related benefit from the
regulation. The agency would need to choose which discount rate to use—and the IWG provides
no guidance on this issue. In this example, the regulation would generate a carbon-related
benefit of $2.62 million, assuming a discount rate of 3%.
B.
Technical Problems Internal to the IWG’s SCC Calculation
There is, to say the least, a great deal of uncertainty surrounding both climate science and
environmental economics. The IWG is admirably modest and forthright about these limitations,
and the last few pages of the document list a number of analytical shortcomings.81 To begin
with, the models do not account for all potential non-catastrophic harms because of data
limitations. For instance, the models do not attempt to quantify the effects of ocean acidification
or the economic and political consequences of population migration that result from warming.82
Nor do they account fully for the possibility of catastrophic harm. For instance, the FUND
model does not account for the possibility of catastrophic harms of any sort, obviously biasing its
estimates of harm downward. The other models may be underestimating the probability of
catastrophic events by significant margins as well.83 In addition, each of the models assumes
81
IWG Report at 5 (“The interagency group offers the new SCC values with all due humility about the uncertainties
embedded in them.”)
82
Id. at 30.
83
Id. at 30-33. There are a number of possible climate change-related “tipping points”: points at which some drastic
climatic event occurs with potentially extreme economic or sociological consequences. These include, for instance,
the collapse of the Greenland Ice Sheet or the West Antarctic Ice Sheet, each of which would raise sea levels
substantially. Climate scientists estimate the likelihood of one of these events occurring at nearly 50%, assuming
2.5 degree global warming. Id. at 33. These “tipping points” would not necessarily generate economic catastrophes
(as they are defined by the climate models), but it is worth noting that the probability of such an event occurring is
nearly an order of magnitude higher than the probability of an economic catastrophe under the DICE or PAGE
models. (PAGE puts the likelihood of a catastrophe at nearly zero, while DICE estimates it at 4%, again assuming
19
that humans will adapt at relatively low cost to many of the results of climate change, though
these assumptions are never justified.84 Indeed, the rate and cost of adaptation is one of the
foremost issues that confounds attempts to estimate the costs of warming. Finally, each of the
models assumes risk neutrality.85 However, people are risk averse, and insurance markets may
not be able to handle the increased risk caused by climate change.86 Rather than calculate a rate
of risk aversion and open itself up to criticism for its choice, the IWG made the more politic (but
less useful) choice to elide the issue entirely.
The IWG is to be commended for acknowledging the weaknesses of its own analysis.
But by the same token, the IWG’s recommendations are not adequately defended. Many of its
errors are likely errors of underestimation: it is likely that the IWG does not incorporate all the
potential harms of global warming, and thus underestimates the benefits of curbing emissions.
But others go the other direction. We describe these additional errors and shortcomings below.
The models use weakly defended assumptions. The models require the input of a set of
economic starting points: estimates of how economic growth and technological development
would proceed absent any global warming. The models are obviously sensitive to these initial
parameters: high economic growth will lead to greater carbon emissions, and thus greater
warming and climate-related harms. In order to accommodate this uncertainty and sensitivity,
the Interagency Group used five different sets of initial parameters, ran each model with each set
of parameters, and then took the average result across all three models under all five sets of
starting points.87 Four of these initial sets of parameters were “business as usual” scenarios—
best estimates of the current state of the economy and its likely growth over the coming
decades.88 One of them was a more optimistic set of parameters—it was actually named
“MERGE Optimistic”—that assumed that technological developments would limit atmospheric
carbon concentrations at relatively low levels for decades into the future.89
The IWG was right to average across a variety of initial economic parameters as a means
of addressing the uncertainty surrounding future economic growth and the sensitivity of climate
models to economic projections. But what could be the justification for choosing four standard
sets of assumptions and one optimistic scenario? Why did the Group not employ one pessimistic
scenario, to balance its use of the optimistic scenario? It is entirely possible that global growth
will drive carbon emissions to levels beyond what standard assumptions would predict. For
instance, if the Chinese economy continues to expand at a high rate and this expansion is fueled
2.5 degree warming.) See also Martin L. Weitzman, On Modeling and Interpreting the Economics of Catastrophic
Climate Change, 91 Rev. Econ. & Statistics 1 (2009); Martin L. Weitzman, A Review on the Economics of Climate
Change, 45 J. Econ. Lit. 703 (2007).
84
IWG Report at 31.
85
Id.
86
Intergovernmental Panel on Climate Change (IPCC), IPPC Fourth Assessment Report Climate Change 2007:
Impacts, Adaptation, and Vulnerability 380-81 (2007).
87
These five sets of economic parameters were borrowed from the Stanford Energy Modeling Forum exercise. Id.
at 16. This modeling exercise developed ten different sets of economic parameters; the Interagency Group selected
five of the ten. Id.
88
Id.
89
Id. at 16-17.
20
primarily through coal and other “dirty” forms of energy production, global carbon emissions
could rise significantly faster than economists and scientists currently predict.90
This raises a larger issue regarding these parameters and the manner in which they are
employed by the models.91 All of the economic parameters, and all of the models, assume that
worldwide carbon intensity—carbon emissions per dollar of GDP—will decline steadily. They
assume that the worldwide economy will become “cleaner” over time, producing fewer and
fewer carbon emissions per dollar of GDP. This is a reasonable assumption for developed
economies that have already industrialized and are in the process of developing cleaner
technologies. But industrializing economies exhibit precisely the opposite behavior: carbon
intensity increases as the economy begins to rely more and more heavily on coal and other
carbon-intensive sources of energy, and only decreases later once a certain level of technology
and prosperity has been achieved. For instance, Chinese carbon intensity rose through the late
1970s, and carbon intensity is still rising in parts of western China that are still industrializing.92
As other nations in Asia, Africa, and South America continue to industrialize in the coming
decades, carbon intensity in those regions may well rise, rather than fall. This will not likely
offset the overall decline in carbon intensity throughout the world, led by technological advances
in the United States, Europe, and China. But it could render assumptions regarding consistent
declines in carbon intensity unrealistic and thereby indicate that the IWG underestimates future
harms due to climate change.
In another important respect the IWG’s analysis may be overly optimistic. None of the
three models incorporates emissions of any greenhouse gas other than carbon dioxide.93 A
regulation that led to lower carbon emissions might simultaneously offset those reductions by
inducing higher emissions of some other greenhouse gas such as methane (an even more potent
greenhouse gas). If firms have the capacity to switch production from some activity that
generates carbon dioxide to another activity that generates methane, the SCC will overestimate
benefits from reductions in carbon emissions.94
The models are crude and inconsistent. At the low levels of emission reductions relevant
here, the DICE, FUND, and PAGE models produce highly divergent results. Recall that the
models are probabilistic; they each produce a range of possible outcomes of varying
probabilities. FUND is the most optimistic: its median estimate of damage from global warming
is lower than all but 5% of the estimates from the PAGE model.95 For its part, the median PAGE
estimate is lower than all but 5% of the estimates from the DICE model. In other words, FUND
Abebe & Masur, supra note.
For discussion, see Daniel A. Farber, Modeling Climate Change and its Impacts: Law, Policy, and Science, 86
Tex. L. Rev. 1655 (2008); see also Daniel A. Farber, Rethinking the Role of Cost-Benefit Analysis, 76 U. Chi. L.
Rev. 1355, 1386-97 (2009) (arguing that cost-benefit analysis cannot handle climate change because of scientific
uncertainty).
92
Abebe & Masur, supra note.
93
IWG Report at 13.
94
This inter-gas leakage is an echo of the even more serious problem of international emissions leakage that we
discuss below.
95
IWG Report at 10. Recall that damages under these models are probabilistic, and thus they produce probability
distributions of possible damages.
90
91
21
estimates are at the extreme low end of PAGE estimates, and PAGE estimates are at the extreme
low end of DICE estimates.
What accounts for these inconsistencies? It is hard to be certain because only the DICE
model is fully transparent. The authors of the other models have not released all of their
equations and calculations to the public (which indeed is an additional reason why it might not
be proper for the government to rely on them). But we can use the DICE model to illustrate a
problem that, we suspect, exists in the other models as well.
The central equation in the DICE model is:
Q(t) = Ω(t)[1-Λ(t)]*A(t)K(t)γL(t)1-γ,
where Q(t) is aggregate global economic output.96 The equation after the “*” is just the standard
neoclassical production function, where A(t) is factor productivity, K(t) is capital stock, and L(t)
is labor inputs. In the neoclassical production function, economic output is a function of capital
investment and labor. Long-run growth is determined by technological innovation and
population changes, which are exogenously assumed. Although neoclassical economic growth
models have received criticism, they are consistent with much empirical evidence, and are
widely used.
DICE’s innovation is to multiply the neoclassical production function by a damage
function, Ω(t)[1-Λ(t)], where Ω(t) maps global mean temperature changes to a damage variable,
and Λ(t) maps abatement expenditures to an abatement variable. As temperature increases, the
damage percentage and abatement costs increase, lowering aggregate economic growth.
Because capital spent on abatement expenses is diverted from capital devoted to economic
growth, aggregate output is maximized when the marginal cost of abatement expenditures equals
the marginal cost of climate-related economic harm. The SCC can be backed out of this model:
it is the price of carbon at which the optimal investment in abatement will be made.
As we noted, the economic growth model is relatively uncontroversial, but where does
the damage function come from? Nordhaus and Boyer explain:
The aggregate damage curve is built up from estimates of the damages of the 12 regions,
including assumed sectoral change and underlying income elasticities of different
outputs. It includes estimated damages to major sectors such as agriculture, the cost of
sea-level rise, adverse impacts on health, and nonmarket damages, as well as estimates of
the potential costs of catastrophic damages. It is clear that this equation is extremely
conjectural, given the thin base of empirical studies on which it rests.97
“Extremely conjectural” is right. The damage function is essentially a guess. And if, as we
suspect, the models differ in large part because their authors chose arbitrarily different damage
96
97
Nordhaus & Boyer, supra at 41.
Nordhaus & Boyer, supra at 41-42.
22
functions, then the SCC just reflects the average of three arbitrary guesses.98 If this is the case, it
is hard to have much confidence in the SCC calculated by the IWG.99
In addition, the manner in which results are reported, coupled with the wide spread
among the three models, conveys an excessively optimistic sense of the potential magnitude of
the climate risk. Recall that the IWG reports the 95th percentile carbon costs (at a 3% discount
rate) as a means of establishing an upper bound on the likely benefits of emissions reductions.100
This 95th percentile figure is an average of the 95th percentile values of all three models. But the
models generate very different 95th percentile predictions—for instance, FUND’s 95th percentile
value is only a small fraction of PAGE’s. Moreover, FUND does not include any potential
catastrophic outcomes. In this sense, the 95th percentile figure is not a realistic bound on the
potential harm from climate change; it is a high estimate mitigated by the much more optimistic
figures produced by FUND. Regulators (and the public) should not be misled into thinking that
there is little chance that carbon costs will exceed the 95th percentile figure reported by the IWG.
To be sure, there is nothing wrong with using a variety of models that make different
predictions. Under conditions of uncertainty, this may be the only wise course of action. But the
extreme discrepancies between these three models—FUND and DICE are essentially
inconsistent with one another—does not inspire confidence. It seems likely that one of the three
models is simply incorrect and is skewing the overall results improperly.101 It was perhaps an
overabundance of caution (both scientific and political) that led the Interagency Group to employ
all three models, rather than attempting to select the most reliable one or two among them. If the
IWG had excluded one or more of these models, it might have opened itself up to criticism for
having cherry-picked available science in order to buttress preferred conclusions. Yet the
inclusion of an inaccurate model is no better than the exclusion of an accurate one.102 In this
respect it appears that the IWG selected what amounts to a political solution to a technical
problem, choosing to eschew difficult technical decisions regarding which models to employ.
This approach ill serves the regulatory agencies who must rely upon the IWG report.
The IWG assumes that the benefits of curbing carbon emissions are linear with the
number of tons of carbon emitted, at least for small reductions in emissions. That is, the IWG
assumes that the benefit of reducing carbon emissions by 20 tons in 2010 is exactly twice the
benefit of reducing emissions by 10 tons in the same year: $420 versus $210 (at a 3% discount
rate).103 This may be a reasonable assumption for very small reductions in national emissions,
98
The models use different economic growth models, but these are unlikely to be the source of much difference,
because all economic growth models attempt to explain the same thing—observed economic growth in different
countries.
99
For a general and somewhat more optimistic discussion of climate modeling, see Daniel A. Farber, Modeling
Climate Change and its Impacts: Law, Policy, and Science, 86 Tex. L. Rev. 1655 (2008).
100
IWG Report at 26.
101
For instance, we find it remarkable that the majority of climate-related damages in North America anticipated by
FUND are in the form of increased demand for air-conditioning. Id. at 8. We are not experts, but this hardly seems
an accurate portrayal of the harm likely to befall farmers and workers in North America in the event of significant
climate change.
102
See Adam M. Samaha, Dead Hand Arguments and Constitutional Interpretation, 108 Colum. L. Rev. 606, 653
(2008) (“Unfortunately, the premises for Condorcetian confidence are lacking. . . . For the Theorem to hold . . . the
mean accuracy of all voters must be better than random--for instance, greater than 50% for binary choices . . . .”).
103
IWG Report at 3.
23
but only for very small reductions. The relationship between the amount of carbon emitted into
the atmosphere and the warming it causes is likely quadratic, cubic, or even exponential, rather
than linear. Similarly, the relationship between global temperature changes and economic harm
is likely nonlinear. Indeed, the DICE and PAGE models assume that the relationship between
temperature change and economic harm obeys a power law: DICE assumes that economic
damages increase as the square of the change in global temperature, while PAGE uses a variety
of values ranging from a linear relationship to an assumption that damages increase as the cube
of temperature changes.104 It is for this reason that the IWG report is unsuitable for measuring
the benefits of major greenhouse gas regulation such as the EPA might undertake in response to
Massachusetts v. EPA. The report itself is explicit on this point.105 If EPA is to calculate the
costs and benefits of across-the-board regulation of carbon dioxide, it must undertake a separate
analysis and cannot rely upon a number of simplifying assumptions made in the course of this
report.
If a regulation causes only a very small reduction in overall carbon emissions, it is
appropriate to approximate these higher-order effects as linear. But it is possible that a number
of independent regulations, in combination, will together reduce emissions by more than a small
fraction. The United States produced approximately 1.65 billion tons of CO2 in 2004.106 Over
the past two years, federal agencies have promulgated a number of regulations that are expected
to cause significant reductions in carbon dioxide emissions, many of them in the millions of tons
annually.107 For instance, the Department of Energy’s July 2009 regulation regarding the energy
efficiency of fluorescent and incandescent lamps is expected to reduce domestic CO2 emissions
by more than 300 million tons over the next thirty years.108 If enough of these regulations were
put into force over the same period of time, they could have nonlinear—quadratic, or even
exponential—effects on global warming. For instance, significant increases in CAFE
standards,109 coupled with improvements in energy efficiency standards for household
appliances110 and related policies, might lead to abatement of emissions that are too great to be
priced accurately using the IWG’s analysis.111 The effects of these small reductions in
combination might be nonlinear, and might be more beneficial than the IWG’s simple model
would indicate. By considering these reductions in piecemeal fashion, agencies might miss
significant interaction effects between regulations. If this is the case, the IWG report would
understate the benefits from reducing carbon emissions by a significant degree.112
104
Id. at 7.
Abebe & Masur, supra note.
See infra Part II.A., infra.
108
See Energy Conservation Program: Energy Conservation Standards and Test Procedures for General Service
Fluorescent Lamps and Incandescent Reflector Lamps, 74 Fed. Reg. 34,080, 34,164 (Table VII.27) (2009); infra
Part II.A.2.
109
E.g., Light-Duty Vehicle Greenhouse Gas Emission Standards and Corporate Average Fuel Economy Standards,
75 Fed. Reg. 25,324 (Envtl. Prot. Agency and Dep’t of Transp. May 7, 2010).
110
Energy Conservation Standards for Certain Consumer Products and for Certain Commercial and Industrial
Equipment, 75 Fed. Reg. 1122-01 (Jan. 8, 2010).
111
See supra Part I.A.1.
112
On the other hand, the relationship could be less than linear, meaning that the Appendix actually overstates
benefits of reducing emissions. However, given the widespread belief among climate modelers that damages scale
more than linearly with increases in global temperature, this seems unlikely.
106
107
24
*
*
*
The IWG report is an admirable attempt to price the benefits of regulatory reductions in
carbon dioxide emissions. Nonetheless, it is rife with uncertainty and error, and internal
disagreements between the models used to generate the final prices do not inspire confidence.
Many of these errors are likely errors of underestimation—the social costs of carbon may well be
higher. Moreover, a consistent theme emerges from these errors: in many cases, the IWG
adopted “political” solutions, designed to appease all sides, where more difficult technical
decisions were called for.
C.
International Leakage
The carbon-absorbing property of the atmosphere is a global public good. People
everywhere in the world benefit when the atmosphere absorbs carbon because the harmful
effects of global warming and other disruptions to the climate are avoided. When one country
limits emissions, and other countries do not, the first country provides benefits to the world,
enjoys only a small share of these benefits, and incurs the full costs of its behavior. The result is
that countries acting unilaterally will reduce carbon emissions inadequately.113
The IWG report recognizes this problem, noting:
Even if the United States were to reduce its greenhouse gas emissions to zero, that step
would be far from enough to avoid substantial climate change. Other countries would
also need to take action to reduce emissions if significant changes in the global climate
are to be avoided. Emphasizing the need for a global solution to a global problem, the
United States has been actively involved in seeking international agreements to reduce
emissions and in encouraging other nations, including emerging major economies, to take
significant steps to reduce emissions. When these considerations are taken as a whole,
the interagency group concluded that a global measure of the benefits from reducing U.S.
emissions is preferable.114
The difficulty lies in the fact that other countries, with the limited exception of the member states
of the European Union, have not taken significant steps to reduce emissions. Thus, the last
sentence does not follow from the sentences that precede it.
To understand why, one must understand the problem of leakage. Leakage occurs when
carbon-emitting activities in one country migrate to another country where regulation of carbon
emissions is weaker. Suppose, for example, that the U.S. government limits carbon emissions of
automobile manufacturers, raising the cost of production, and hence the price of automobiles,
reducing the demand for American-made cars. All else equal, American consumers will, at the
margin, stop buying American-made cars and instead buy foreign-made cars that are not subject
For recent discussions in the law review literature, see Michael P. Vandenbergh & Mark A. Cohen, Climate
Change Governance: Boundaries and Leakage, 18 NYU Envt’l L.J. 221 (2010); Robert N. Stavins, A Meaningful
U.S. Cap-and-Trade System to Address Climate Change, 32 Harv. Envtl. L. Rev. 293 (2008); Jonathan B. Wiener,
Climate Change Policy, and Policy Change in China, 55 UCLA L. Rev. 1805 (2008).
114
IWG Report at 11-12.
113
25
to the American-imposed carbon cap. As a result, foreign automobile manufacturers will expand
production and hence generate greater carbon emissions—potentially offsetting or even
exceeding the reduction in carbon emissions that took place in the United States.115 Because the
harm to the climate is independent of the source of the emissions, the American cap will have no
effect on climate change. Its only effect will be job loss in the short term, and, in the long term,
a transfer of carbon-intensive industry from American soil to foreign soil, which may well be
less economically efficient (as it requires transportation of cars back to the United States, loss of
comparative advantage efficiencies, and so on).116
Consider another way in which leakage occurs. Suppose that American climate
regulation increases the cost of oil for American industry and consumers, resulting in a decline in
demand. Because American demand for oil declines, and oil is bought and sold on a global
market, the global price of oil will decline. As the global price of oil declines, people in other
countries will buy more of it. Depending on the precise shape of the supply and demand curves,
the quantity of oil consumed in the new equilibrium may be close to the quantity of oil consumed
prior to the introduction of American climate regulation. Foreign-produced oil originally
intended for the American market will be sent to other countries, outside of American regulatory
control.
In this extreme scenario, leakage is one hundred percent, with the result that the
American regulation produces no social gains and only costs. However, there are two constraints
on leakage. The first might be described as physical or technological. In our first example, it
may turn out to be the case that the increase in the price of cars will have a limited effect on
demand (demand is relatively inelastic). Perhaps, for example, the cost of transporting cars from
overseas is very high; or American automobile manufacturers simply produce more desirable
cars.117 It is also possible that foreign manufacturing processes are less carbon-intensive than
American manufacturing processes.118 In any of these cases, the reduction in carbon emissions
from American manufacturers will only be partly offset—say 10 or 30 or 70 percent—by the
increase in carbon emissions from foreign manufacturers. The extent of these offsets will vary
from industry to industry.
The second constraint is legal. Suppose the U.S. government enacts a law that bans or
taxes the importation of all foreign automobiles, or those that were produced using excessively
115
Foreign production of automobiles might increase carbon emissions because energy is “dirtier” in, for instance,
China than it is in the United States. Because Chinese electricity producers rely more heavily on coal and have
installed fewer pollution-controlling technologies than American producers, each unit of energy produced in China
generates correspondingly greater quantities of carbon emissions. Abebe & Masur, supra.
116
For a discussion, see, e.g., Mustafa H. Babiker, Climate Change Policy, Market Structure, and Carbon Leakage,
65 J. Intern’l Econ. 421 (2005).
117
Neither of these possibilities appears to be true, but they could be in other industrial contexts.
118
This is likely true with respect to some countries, but not with respect to others. The United States ranks 92nd
among the 133 nations (for which there are data) in the carbon intensity of its energy production. That is to say, 91
countries produce less carbon per unit of energy than the United States; 41 produce more. World Resources
Institute, Climate Analysis Indicators Tool, Carbon Intensity of Energy Use in 2006,
available at
http://cait.wri.org/cait.php?page=carberg&mode=view. Notably, China, Australia, Ireland, Israel, Italy, and
Denmark have less carbon-efficient economies than the United States; Japan, India, Canada, and most European
countries are more efficient. Id. Though energy is not the only input to production, the carbon efficiency of energy
production is roughly correlated with the carbon efficiency of production more generally.
26
carbon-intensive manufacturing processes. Such a law would require foreign producers either to
abandon the American market or to produce less carbon-intensive cars for export to the United
States. In either case, the leakage problem would be mitigated. In reality, laws that address
leakage by restricting trade would have to overcome a number of hurdles, including international
trade law, and the difficulty of determining the carbon intensity of particular goods, and it is
highly unlikely that they could eliminate leakage.119 Because these laws do not yet exist,
agencies must take into account leakage.
Scholars have produced a number of estimates of leakage. At this stage, the estimates are
highly speculative. A study by Joshua Elliott and his coauthors considers the problem of leakage
in a scenario in which all Annex B countries in the Kyoto treaty (major developed nations) adopt
a carbon tax and the non-Annex B countries do not.120 Using a new computable general
equilibrium model of the economic effects of climate change, they find that a regional tax would
reduce emissions only one third as much as an equivalent global tax, and that the effectiveness of
the regional tax would be reduced a further 15 to 25 percent as a result of leakage. Constraints
on emissions in the European Union, the United States, and other Annex B countries would
increase production and consumption of carbon-intensive goods by 15 to 25 percent in China,
India, and other non-Annex B countries.
These figures are suggestive of the problem of leakage if climate regulation were to occur
in the United States alone rather than in all Annex B countries. Leakage would now occur not
only in China and India, but also in the European Union (the largest economy in the world),
Japan (the second largest national economy), Canada (the tenth largest), and Australia (the
thirteenth largest). To be sure, the European Union already has a system of carbon regulation,
which would limit leakage to some extent, but there is some question how effective that system
actually is.121 Overall, leakage in a scenario in which the United States alone (or the United
States and the European Union) engage in carbon regulation, and other countries do not, would
certainly be higher than 15 to 25 percent.
To account for this problem, agencies must incorporate leakage estimates into their costbenefit analyses involving the SCC. Consider a regulation that reduces carbon emissions by one
metric ton. Using the IWG’s figure for a discount rate of 3 percent, the benefit of the regulation
would be $21.40. If we conservatively assume that leakage is 25 percent, this means that the one
ton reduction of carbon emissions will be offset by a quarter ton rise of carbon emissions in
another part of the world. Accordingly, either the agency should discount the expected carbon
emission by one quarter, or discount the SCC by one quarter. In either event, the benefit of the
regulation would be reduced to $16.05.
Unfortunately, the calculation would not be this simple. The Elliott et al. study i comes
to results that conflict with other studies that estimate leakage rates from as little as 3 percent to
See Paul-Erik Veel, Carbon Tariffs and the WTO: An Evaluation of Feasible Policies, 12 J. Int’l Econ L. 749,
755-60 (2009).
120
Joshua Elliott, Ian Foster, Sam Kortum, Todd Munson, Fernando Pérez Cervantes, & David Weisbach, A
Quantitative Examination of Trade and Carbon Taxes (unpub. m.s., 2010).
121
See Frank Convery, Denny Ellerman, and Christian De Perthuis, The European Carbon Market in Action:
Lessons
From
the
First
Trading
Period
(2008),
available
at
http://mit.dspace.org/bitstream/handle/1721.1/44619/MITJPSPGC_Rpt162.pdf?sequence=1.
119
27
more than 100 percent.122 None of these studies examine the specific case in question where the
United States but no other country (or only the European Union) adopt significant constrains on
carbon emissions. The studies also address different types of leakage—for example, leakage that
results from regulation of coal versus leakage that results from regulation of petroleum, which
need not be the same. Agencies that regulate products like appliances that draw on coal as their
energy source must use leakage numbers different from those used by agents that regulate
products like mortor vehicles that draw on petroleum as their energy source. Petroleum is traded
on global markets to a much greater degree than coal is, and thus susceptible to greater leakage.
Similarly, regulations concerning methods of production, such as the production of steel, are
much more susceptible to leakage than regulations concerning domestic usage, such as
regulations of the efficiency of automobiles123 or small electric motors.124 Production can be
transferred overseas to evade regulation, while usage (by Americans) cannot. Further research
into leakage rates will be necessary before the SCC can be appropriately discounted.
In sum, leakage poses two problems. First, the degree of leakage in current conditions is
unknown. Second, and more important, cost-benefit analyses that ignore leakage will
overestimate the benefits of a regulation. The amount of leakage will eventually depend on
international cooperation, and in particular a climate treaty. How might agencies currently take
account of this? It would be wrong for agencies to assume that leakage is zero or very low
because a climate treaty that eliminates leakage will eventually come into existence. A climate
treaty might or might not eventually come into existence. But if agencies assume that leakage is
very high, they will rarely engage in climate regulation. We return to this dilemma in Part III.
D.
Valuation of Harms to Foreign Countries
Climate regulation is distinctive because optimal domestic climate regulation depends on
the climate regulation of other countries. This is not true for virtually all other forms of
regulation. Consider regulation of arsenic in municipal water supplies.125 The adverse effect of
arsenic is felt overwhelmingly by Americans; the cost of remedial measures also shows up in the
water bills paid by Americans living in the cities affected by the regulation. Aside from a few
tourists and other foreign visitors, the costs and benefits are felt by Americans.
Joseph Aldy & William A Pizer, The Competitiveness Impacts of Climate Change Mitigation Policies, The Pew
Center on Global Climate Change (2009) (noting problem of leakage); Babiker, supra (estimating leakage rates as
high as 130 percent and as low as 50 percent for countries required to reduce emissions by the Kyoto Protocol);
Mustafa H.M. Babiker & T.F. Rutherford, The Economic Effects of Border Measures in Subglobal Climate
Agreements, 26 Energy J. 99, 106-08 (2005) (estimating leakage from 7 percent to 39 percent dependant on whether
the U.S. is included in the Kyoto coalition countries and on what type of protective measures, such as tariffs or
exemptions, Kyoto coalition countries take to reduce leakage); Ton Manders & Paul Veenendaal, Border Tax
Adjustments and the EU-ETS: A Quantitative Assessment, CPB Documents 171 (2008) (estimating a 3.3 percent
leakage rate for reductions made by the European Union).
123
See Average Fuel Economy Standards Passenger Cars and Light Trucks Model Year 2011, 74 Fed. Reg. at
14,196 (Dep’t of Transportation March 30, 2009).
122
124
Energy Conservation Program: Energy Conservation Standards for Small Electric Motors, 75
Fed. Reg. 10,874 (Dep’t of Energy March 9, 2010).
See, e.g., National Primary Drinking Water Regulations, 66 Fed. Reg. 6976, 6981 (Jan. 22, 2001) (codified at 40
C.F.R. pts. 9, 141, 142).
125
28
By contrast, when American industry reduces carbon emissions, the beneficiaries are
mainly foreigners. The population of the United States is less than five percent of the global
population; and future population growth will occur disproportionately outside the United States.
Accordingly, reduction of carbon emissions today will benefit far more foreigners, living today
and in the future, than Americans. Meanwhile, the costs will be felt mostly although not
exclusively by Americans. Most of the costs will be passed on to consumers, who are mostly
Americans. Market share will be lost in both domestic and foreign markets because foreign
manufacturers do not face the same regulations.
Thus, we can distinguish the global SCC and the domestic SCC. When U.S. industries
emit an additional ton of carbon dioxide, the climate-related harm will be partly felt by
Americans and entirely felt by the global population (including Americans). For example, if
global food prices increase as a result of harm to agricultural productivity, the reduction in
wealth will be felt mostly outside the United States, where 95 percent of the global population is
located. Accordingly, the domestic SCC will be less than the global SCC. A cost-benefit
analysis that takes into account the well-being only of Americans will use the domestic SCC; a
cost-benefit analysis that takes into account the well-being of the global population will use the
global SCC. A cost-benefit analysis using a global SCC will yield more stringent regulations
than cost-benefit analysis using a domestic SCC.
Should agencies such as the EPA use the global SCC or the domestic SCC in its costbenefit analysis? The legal answer is that it depends on the statute. However, statutes are almost
always silent about extraterritorial effect. Consider the statute under which EPA may end up
regulating climate emissions:
The Administrator shall by regulation prescribe (and from time to time revise) in
accordance with the provisions of this section, standards applicable to the emission of any
air pollutant from any class or classes of new motor vehicles or new motor vehicle
engines, which in his judgment cause, or contribute to, air pollution which may
reasonably be anticipated to endanger public health or welfare . . . .126
The term, “air pollutant,” means “any air pollution agent or combination of such agents,
including any physical, chemical, biological, radioactive . . . substance or matter which is emitted
into or otherwise enters the ambient air.”127
The question is whether the relevant “public” in the statute includes only Americans or
the global population. In most statutes, two considerations would suggest the narrower
interpretation.
First, courts frequently apply a canon of interpretation that disfavors
extraterritorial application of laws.128 The presumption is that Congress has no interest in
regulating overseas—because doing so may offend the sovereignty of other countries, while
producing few benefits for Americans. Second, the contrary interpretation implies that EPA
should regulate, or attempt to regulate, foreign motor vehicle manufacturers, even those who
126
42 U.S.C. § 7521(a)(1).
42 U.S.C. § 7602(g).
128
Recently affirmed by the Supreme Court in Morrison v. National Australia Bank, 2010 WL 2518523 (U.S.
2010).
127
29
produce only for their domestic markets. Such an interpretation is questionable, to say the
least.129
However, the Clean Air Act has a clause that provides for extraterritorial effect with
respect to countries that reciprocate by regulating their domestic industry so as to benefit the
United States.130 This suggests that neither the domestic nor global SCC should be used. The
SCC should be calculated so as to take into account impacts on countries that enter a climate
treaty or otherwise engage in climate regulation, and no others. Notice how far removed this
approach is from cost-benefit analysis, which does not limit benefits on the basis of the political
behavior of governments.
The fourteen regulations that we have been discussing were not issued under statutes with
reciprocity provisions. Yet some of them distinguish domestic and global SCCs. As we noted in
Part I.A., the Model Year 2011 CAFE rule began with a $33 price for the global SCC. It then
separately calculated a domestic SCC on the basis of the U.S. share of world economic output,
about 20 to 28 percent, and estimates of the sensitivity of the U.S. economy to climate change.
According to this calculation, the United States would suffer between 0 and 14 percent of the
global cost of climate change. DOT took the midpoint, 7 percent, and multiplied it by the $33
global SCC, yielding a domestic SCC of approximately $2.131
The calculation is dubious. The cost of climate change is not a function of the size of a
country’s economy, but its geography (for example, whether it is low-lying and subject to
flooding), its mean temperatures (hotter places suffer more from increased heat than colder
places), the health of the population (healthier people are less susceptible to disease, which
increases in warmer climates), and so forth. Scientists have not yet been able to predict regional
variation in the effects of climate change with any confidence, and for this reason the major
climate models do not yet make predictions for individual countries—although some economists
have tried.132 Speculations as to the sensitivity of the American economy are merely that. It
follows that even if a global SCC can be calculated, a domestic SCC cannot be estimated to any
reasonable degree of confidence—although we can be confident that it is less than the global
SCC.
Even if American regulatory agencies have the authority to take account of costs to
foreigners, should they use their discretion to act in that way? This question can be broken down
into three inquiries—normative, political, and institutional.
The normative question is whether Americans have the same moral obligations to
foreigners who live overseas as they have toward other Americans. If Americans have weaker
129
For a recent explanation for the presumption of extraterritoriality in litigation involving environmental laws, see
Jonathan Remy Nash, The Curious Legal Landscape of the Extraterritoriality of U.S. Environmental Laws, 50 Va. J.
Int’l L. 997 (2010).
130
42 U.S.C. § 7415(c).
131
See supra.
132
See, e.g., Mark A. Delucchi, Summary of the Non-Monetary Externalities of Motor Vehicle Use, UCD-ITS-RR96-3 (9) rev. 1, Institute of Transportation Studies, University of California, Davis, originally published September
1998, revised October 2004, pp. 49-51. Available at http://www.its.ucdavis.edu/publications/2004/UCD-ITS-RR-9603(09)_rev1.pdf.
30
moral obligations to foreigners, then American regulatory agencies should discount (partially or
completely) the negative effects of American activities on foreigners. Philosophers take
different positions on this question. Cosmopolitans believe that moral obligations to help or not
to harm others do not turn on their nationality; communitarians and nationalists believe that they
do.133 We do not take a position on this debate; we note only that if Americans owe stronger
duties to other Americans than to foreigners, then the social cost of carbon should reflect a
discount for costs incurred by foreigners. And even if Americans should treat foreigners the
same as other Americans, there is still a question whether to incorporate foreigners’ own
valuations, which, in fact, differ from those of Americans, or to use American valuations—that
is, to value foreigners well-being the same as Americans value their own well-being.134
Suppose, for example, that Chinese and Italians assign a lower value to morbidity and mortality
risks than Americans. Should American regulatory agencies use Chinese and Italian valuations
or American valuations for climate effects on Chinese and Italians?135
The political question is whether elected officials and appointees who head regulatory
agencies should treat costs to foreigners the same as costs to Americans even if they believe it is
appropriate to do so—or, as a practical matter, whether they can be expected to do so in a
democracy. Suppose that most elected officials are cosmopolitans but most Americans are
nationalists. For instance, one study suggests that Americans implicitly value a foreign life at
between one sixth and 1/2000th of an American life.136 Even if we should all be cosmopolitans,
it is plausible to believe that elected officials have a duty to advance the interests of the
Americans they represent, as do the regulatory agencies that derive their authority from the
decisions of elected officials. In a more practical sense, electoral pressures may dictate that
governments treat foreigners, who do not have the vote, differently from the way they treat
citizens.
The institutional question addresses how regulatory agencies should act, given that they
have specialized missions. Even if the cosmopolitan view is correct, and even if politicians
believe that they should take account of the interests of foreigners, it may well be the case that
regulatory agencies should not—unless explicitly directed to do so by elected officials.
There are two reasons for doubting that agencies should treat costs to foreigners and costs
to Americans as the same. First, the treatment of foreigners is a diplomatic question, one that is
addressed by different institutions such as the state department, the White House, and sometimes
Congress. To see the problem here, imagine that the optimal approach to climate change is joint
action by multiple countries, as is surely the case. The United States and other countries do best
if all countries reduce greenhouse gas emissions by a reasonable amount.137 However, countries
disagree about how significant climate change is, and how the costs of responding to climate
133
See generally Simon Caney, Justice Beyond Borders: A Global Theory 3-16 (2005) (describing competing
approaches to international relations including the cosmopolitan, realist, statist, and nationalist approaches).
134
See W. Kip Viscusi and Joseph E. Aldy, The Value of a Statistical Life: A Critical Review of Market Estimates
Throughout the World, 27 J Risk & Uncertainty 5, 26-30 & tble. 4 (2003) (showing a range of valuations for a
statistical life from $0.8 million in South Korea to $19.9 million in the United Kingdom, in 2000 U.S. dollars).
135
See Eric A. Posner & Cass R. Sunstein, Dollars and Death, 72 U. Chi. L. Rev. 573 (2005).
136
Wojciech Kopczuk, Joel Slemrod, and Shlomo Yitzhaki, The Limitations of Decentralized World Redistribution:
An Optimal Taxation Approach, 49 European Econ. Rev. 1051, 1075 (2005).
137
See Cass Sunstein, Of Montreal and Kyoto: A Tale of Two Protocols, 31 Harv. Envt’l. L. Rev. 1 (2007).
31
change should be allocated among countries. Countries must resolve these disagreements—in
effect, choosing a global SCC—before they can cooperate.
It is possible that unilateral action by the United States will induce reciprocal behavior by
other countries. But the opposite is more probable: if the United States reduces greenhouse gas
emissions unilaterally, other countries will fail to reciprocate, preferring to free ride.138 Suppose
the United States can compel other countries to reduce greenhouse gas emissions only by
threatening not to reduce greenhouse gas emissions itself. If so, then it would be wrong for
American regulatory agencies to reduce carbon emissions in a way that reflects the interests of
foreigners until directed to do so by the political branches.
In the years leading up to the Montreal Protocol, which restricts the use of
chlorofluorocarbons and other compounds that cause stratospheric ozone depletion, the United
States acted first, issuing regulations that greatly reduced use of these compounds by consumers
and domestic manufacturers. When it came time to negotiate a treaty with other countries, the
United States had difficulty persuading them to take similar steps. These countries may have
preferred to free ride on American efforts, which conferred benefits on them without imposing
costs. The final treaty was less stringent than the United States sought.139 It is possible that this
disappointment resulted from the self-imposed reduction in bargaining power on the part of the
United States, although one cannot know for sure.
Second, the treatment of foreigners is a highly sensitive political question. As noted
above, it is not clear whether elected officials should take account of foreign interests to the same
extent as domestic interests. And as a practical matter, elected officials must pay attention to the
views of voters. Congress is the institutional actor best positioned to decide these normatively
and politically fraught questions. If Congress will not act, then the administration should at least
employ notice-and-comment rulemaking in order to increase the degree of political
accountability and participation that attends the decision.140 An interagency regulatory group,
working behind closed doors with no public notice or participation, is not the proper mechanism
for resolving such issues.
III. The Limits of Cost-Benefit Analysis
Many of the problems in the IWG report are technical. The SCC chosen by the IWG
ignores some significant costs, neglects catastrophic risks, and assumes away risk aversion.
Because the figure assumes a linear relationship between the benefits of carbon reduction and the
quantity of carbon reduction, it cannot be used for large-scale regulations or even small-scale
regulations if they aggregate to a large scale. The SCC reflects arbitrarily chosen damage
functions (or, more precisely, the average of several arbitrarily chosen damage functions).141 It
emerges from three climate models of questionable reliability.
See Rachel Brewster, Stepping Stone or Stumbling Block: Incrementalism and National Climate Change
Legislation 28 Yale L. & Pol’y Rev. 245 (2010) (arguing that incremental domestic action may retard the
development of an international agreement on climate change).
139
See id. at 9-17 (describing U.S. domestic regulation and international negotiations).
140
Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245 (2001).
141
This is in addition to the fact that scientific uncertainty makes cost-benefit analysis difficult. See supra Part II.B.
138
32
Guesswork is not always fatal to cost-benefit analysis. Judgment is needed to distinguish
between reasonable estimates and estimates that are excessively wide of the mark. Our judgment
is that the IWG report relies on too many unwarranted assumptions and cannot be relied on. But
technical problems can be repaired through further research. Computer models of climate
change are improving and will eventually provide reasonable estimates of the SCC.
The more serious problem for cost-benefit analysis of climate change is that climate
regulation requires a series of judgments that are political rather than technocratic. We review
those judgments here.
First, optimal U.S. climate regulation depends on what other countries do. If other
countries do not engage in their own climate regulation, then U.S. climate regulation should
probably be less strict, given the possibility of leakage. If other countries do engage in their own
climate regulation, then U.S. regulators need not fear leakage as much and can regulate carbon
more strictly. A given regulation may be cost-justified if China also acts but not if China fails to
act. Yet cost-benefit analysis does not permit for speculation about how other countries will
behave. Agencies such as the DOT and DOE are in no position to predict whether China will
regulate carbon or not, and so cannot, consistent with the norms of cost-benefit analysis, use an
SCC that assumes action on the part of the China.
Second, climate regulation by the United States might itself affect how other countries
act. There are two possibilities. One is that if the United States moves first, it provides
leadership that motivates other countries to engage in climate regulation. The other is that if the
United States moves first, other countries free ride and refuse to engage in climate regulation.
Cost-benefit analysis cannot predict the likely diplomatic consequences of agency actions.
Third, U.S. climate regulation will affect the well-being of many more foreigners than
Americans. Yet there is no agreed-upon formula for evaluating the benefits of avoided harms for
non-Americans living in other countries. An agency might use American valuations; or it might
defer to the valuations used in foreign countries; or it might exclude foreign costs and benefits.
The right approach depends on normative questions that are prior to cost-benefit analysis, and
institutional and political questions that cost-benefit analysis does not address.
Fourth, although a scientific consensus holds that anthropogenic climate change will
cause significant harm to people around the world, there is no scientific consensus regarding how
much of that harm will be experienced by Americans in the United States. If cost-benefit
properly takes into account only effects in the United States, then agencies should engage in only
limited climate regulation.
The broader point is that cost-benefit analysis will be ineffective whenever a regulation
raises principally normative and political questions, rather than technocratic ones. This is why
cost-benefit analysis is considered inapplicable to abortion and Establishment Clause questions,
for instance. Without some mechanism for evaluating the normative and political questions
central to those subjects, cost-benefit analysis has little purchase.
33
We suspect that all of these difficulties have caused agencies to avoid relying on a single
SCC in their regulatory impact statements, for fear that a court would find the cost-benefit
analysis defective because the SCC figures are unsupported. However, ignoring these problems
does not solve them. As we noted earlier, the Ninth Circuit struck down an earlier CAFE rule
because DOT failed to take into account the SCC.142
Yet if the IWG’s cost-benefit analysis is not suited to the task of pricing carbon, what
should the government do? In an ideal world, the United States and other countries would enter
into a climate treaty that would establish the SCC. Note that this SCC would be a political
figure, reflecting a compromise that different nations reach for political purposes, rather than a
figure determined using traditional economic methodology. But the figure could be easily
plugged into cost-benefit analyses used by agencies for regulation—at least, as long as agencies
are given proper executive or congressional authorization to use a politically stipulated figure
rather than the actual figure that emerges from climate models. A climate treaty could also
render agency regulation unnecessary by creating a cap-and-trade scheme. Because people
would have to pay for the harm they cause to the climate, they would fully internalize the
climatic costs of their activity, and additional regulation administered by the regulatory agencies
would be unnecessary. Congress would need to pass a law that provides that the government
must treat the treaty-based SCC as if it were the real social cost of carbon, to ensure that
additional regulations beyond American treaty commitments are not issued.
However, a climate treaty is still far off; what should U.S. regulatory agencies do in the
meantime? We believe that agencies conducting cost-benefit analysis cannot use the IWG’s
SCC. The SCC is highly arbitrary: it assumes that foreign harms should be treated identically to
domestic harms; assumes that there will be zero leakage; and assumes away the possibility of a
diplomatic solution. Even the choice of which of the IWG’s four SCCs to use is arbitrary. (It
may well be for this reason that agencies are at such pains to argue that the choice of SCC will
not affect the regulatory decision.) As a matter of policy and law, cost-benefit analysis cannot
use arbitrary valuations.143 It is possible that a case could be made for a much lower SCC, using
lower bound estimates for American impacts that are scientifically supportable, or even for a
higher SCC. But we think that the more honest and plausible approach would be for Congress to
stipulate an SCC that reflects its political judgments (for example, that a high SCC would
encourage rather than discourage other countries to join a climate treaty). If Congress will not
act, then the Obama administration should suspend Executive Order 12,296 for regulations
touching on climate change and order agencies to use a figure that will encourage other countries
to enter a climate treaty (if in fact such a figure exists);144 and initiate notice-and-comment
rulemaking as a second-best means of addressing the political questions that cost-benefit analysis
cannot answer.
142
Ctr. for Biological Diversity, 538 F.3d 1172.
See 5 U.S.C. § 706(2)(A) (ordering courts to set aside agency action that is “arbitrary [or] capricious”).
144
Whether such an order would be lawful would depend on the proper interpretation of the particular statutes
involved, taking into account the conflicting pressures of the Chevron doctrine, which confers discretion on the
executive to achieve policy objectives, and the presumption against extraterritoriality.
143
34
Conclusion
Contrary to popular perception, the U.S. government has already begun engaging in
climate regulation. Congressional paralysis on climate legislation has not stood in the way
because U.S. regulatory agencies derive authority from their authorizing statutes and the costbenefit executive order, which requires them to take into account all the possible benefits of
regulation.
That is the good news. The bad news is that the agencies’ regulatory efforts have been
inadequate. Simultaneously miscalculating the SCC and ignoring their own numbers, agencies
manage to do cost-benefit analysis poorly and then disregard it. Much improvement can take
place through purely technical adjustment, and the IWG has led the way. Many of the problems
that we have identified can be resolved through further research or reasonable guesswork. But it
is important to understand that although more precise domestic and global SCC figures can be
useful information for political actors, they cannot be used in a conventional cost-benefit analysis
performed at the agency level because serious political issues remain as a result of the global
nature of climate change and the uncertainties that continue to surround it. For this reason, the
cost-benefit executive order does not provide a foundation for further agency regulation.
Congressional or executive intervention is necessary.
Readers with comments should address them to:
Professor Jonathan S. Masur
University of Chicago Law School
1111 East 60th Street
Chicago, IL 60637
[email protected]
35
Chicago Working Papers in Law and Economics
(Second Series)
For a listing of papers 1–450 please go to Working Papers at http://www.law.uchicago.edu/Lawecon/index.html
451.
452.
453.
454.
455.
456.
457.
458.
459.
460.
461.
462.
463.
464.
465.
466.
467.
468.
469.
470.
471.
472.
473.
474.
475.
476.
477.
478.
479.
480.
481.
482.
483.
484.
485.
486.
487.
488.
489.
Tom Ginsburg, Public Choice and Constitutional Design (January 2009)
Richard Epstein, The Case against the Employee Free Choice Act (January 2009)
Adam B. Cox, Immigration Law’s Organizing Principles (February 2009)
Philip J. Cook, Jens Ludwig, and Adam M. Samaha, Gun Control after Heller: Threats and Sideshows from
a Social Welfare Perspective (February 2009)
Lior Jacob Strahilevitz, The Right to Abandon (February 2009)
M. Todd Henderson, The Nanny Corporation and the Market for Paternalism (February 2009)
Lee Anne Fennell, Commons, Anticommons, Semicommons (February 2009)
Richard A. Epstein and M. Todd Henderson, Marking to Market: Can Accounting Rules Shake the
Foundations of Capitalism? (April 2009)
Eric A. Posner and Luigi Zingales, The Housing Crisis and Bankruptcy Reform: The Prepackaged Chapter
13 Approach (April 2009)
Stephen J. Choi, G. Mitu Gulati, and Eric A. Posner, Are Judges Overpaid? A Skeptical Response to the
Judicial Salary Debate (April 2009)
Adam B. Cox and Eric A. Posner, The Rights of Migrants (April 2009)
Randal C. Picker, The Google Book Search Settlement: A New Orphan-Works Monopoly? (April 2009,
revised July 2009)
Randal C. Picker, The Mediated Book (May 2009)
Anupam Chander, Corporate Law’s Distributive Design (June 2009)
Anupam Chander, Trade 2.0 (June 2009)
Lee Epstein, William M. Landes, and Richard A. Posner, Inferring the Winning Party in the Supreme Court
from the Pattern of Questioning at Oral Argument (June 2009)
Eric A Posner, Kathryn Spier, and Adrian Vermeule, Divide and Conquer (June 2009)
John Bronsteen, Christopher J. Buccafucso, and Jonathan S. Masur, Welfare as Happiness (June 2009)
Richard A. Epstein and Amanda M. Rose, The Regulation of Soveriegn Wealth Funds: The Virtues of
Going Slow (June 2009)
Douglas G. Baird and Robert K. Rasmussen, Anti-Bankruptcy (June 2009)
Bernard E. Harcourt, Alon Harel, Ken Levy, Michael M. O’Hear, and Alice Ristroph, Randomization in
Criminal Justice: A Criminal Law Conversation (June 2009)
Bernard E. Harcourt, Neoliberal Penality: A Brief Genealogy (June 2009)
Lee Anne Fennell, Willpower and Legal Policy (June 2009)
Richard A. Epstein, How to Undermine Tax Increment Financing: The Lessons of ProLogis v. City of
Chicago (June 2009)
Randal C. Picker, Online Advertising, Identity and Privacy (June 2009)
M. Todd Henderson, Credit Derivatives Are Not “Insurance” (July 2009)
Lee Anne Fennell and Julie Roin, Controlling Residential Stakes (July 2009)
Douglas G. Baird, The Holmesian Bad Man’s First Critic (August 2009)
Douglas G. Baird, The Bankruptcy Exchange (August 2009)
Jonathan Masur and Eric A. Posner, Against Feasibility Analysis (August 2009)
Lee Anne Fennell, The Unbounded Home, Property Values beyond Property Lines (August 2009)
Bernard E. Harcourt, Henry Louis Gates and Racial Profiling: What’s the Problem? (September 2009)
Stephen J. Choi, Mitu Gulati, Mirya Holman, and Eric A. Posner, Judging Women (September 2009)
Omri Ben-Shahar, One-Way Contracts: Consumer Protection without Law (October 2009)
Ariel Porat, Expanding Liability for Negligence Per Se (October 2009)
Ariel Porat and Alex Stein, Liability for Future Harm (October 2009)
Anup Malani and Ramanan Laxminrayan, Incentives for Surveillance of Infectious Disease Outbreaks
(October 2009)
Anup Malani, Oliver Bembom and Mark van der Laan, Accounting for Differences among Patients in the
FDA Approval Process (October 2009)
David Gilo and Ariel Porat, Viewing Unconsconability through a Market Lens (October 2009)
36
490.
491.
492.
493.
494.
495.
496.
497.
498.
499.
500.
501.
502.
503.
504.
505.
506.
507.
508.
509.
510.
511.
512.
513.
514.
515.
516.
517.
518.
519.
520.
521.
522.
523.
524.
525.
David Weisbach, Instrument Choice Is Instrument Design (October 2009)
M. Todd Henderson, Justifying Jones (November 2009)
Eric A. Posner, ProCD v. Zeidenberg and Cognitive Overload in Contractual Bargaining (November 2009)
Randal C. Picker, Antitrust and Innovation: Framing Baselines in the Google Book Search Settlement
(November 2009)
Richard A. Epstein, Against Permititis: Why Volunteer Organizations Should Regulate the Use of Cancer
Drugs (November 2009)
Richard A. Epstein, Heller’s Gridlock Economy in Perspective: Why There Is Too Little, Not Too Much,
Private Property (November 2009)
Richard A. Epstein, NRA v. City of Chicago: Does the Second Amendment Bind Frank Easterbrook?
(November 2009)
Randal C. Picker, Easterbrook on Copyright (November 2009)
Omri Ben-Shahar, Pre-Closing Liability (November 2009)
Randal C. Picker, Assessing Competition Issues in the Amended Google Book Search Settlement
(November 2009)
Saul Levmore, Ambigious Statutes (November 2009)
Saul Levmore, Interest Groups and the Problem with Incrementalism (November 2009)
Tom Ginsburg, The Arbitrator as Agent: Why Deferential Review Is Not Always Pro-Arbitration
(December 2009)
Nuno Garoupa and Tom Ginsburg, Reputation, Information and the Organization of the Judiciary
(December 2009)
Eric A. Posner and Alan O. Sykes, Economic Foundations of the Law of the Sea (December 2009)
Jacob E. Gersen and Anne Joseph O’Connell, Hiding in Plain Sight? Timing and Transparency in the
Administrative State (December 2009)
Richard A. Epstein, Impermissible Ratemaking in Health-Insurance Reform: Why the Reid Bill is
Unconstitutional (December 2009)
Tom Ginsburg and Eric A. Posner, Subconstitutionalism (January 2010)
Stephen J. Choi, Mitu Gulati, and Eric A. Posner, What Do Federal District Judges Want? An Analysis of
Publications, Citations, and Reversals (January 2010)
Joseph Isenbergh, The Future of Taxation (January 2010)
Lee Epstein, William M. Landes, and Richard A. Posner, Why (and When) Judges Dissent: A Theoretical
and Empirical Analysis (January 2010)
Tom Ginsburg, James Melton, and Zachary Elkiins, The Endurance of National Constitutions (February
2010)
Omri Ben-Shahar and Anu Bradford, The Economics of Climate Enforcement (February 2010)
Neta-li E. Gottlieb, Free to Air? Legal Protection for TV Program Formats (February 2010)
Omri Ben-Shahar and Eric A. Posner, The Right to Withdraw in Contract Law (March 2010)
Richard A. Epstein, Inside the Coasean Firm: Competence as a Random Variable (March 2010)
Omri Ben-Shahar and Carl E. Schneider, The Failure of Mandated Disclosure (March 2010)
Kenneth W. Dam, The Subprime Crisis and Financial Regulation: International and Comparative
Perspectives (March 2010)
Lee Anne Fennell, Unbundling Risk (April 2010)
Stephen J. Choi, Mitu Gulati, and Eric A. Posner, Judicial Ability and Securities Class Actions (April 2010)
Jonathan S. Masur and Jonathan Remy Nash, The Institutional Dynamics of Transition Relief (April 2010)
M. Todd Henderson, Implicit Compensation, May 2010
Lee Anne Fennell, Possession Puzzles, June 2010
Randal C. Picker, Organizing Competition and Cooperation after American Needle, June 2010
Richard A. Epstein, What Is So Special about Intangible Property? The Case for intelligent Carryovers,
August 2010
Jonathan S. Masur and Eric A. Posner, Climate Regulation and the Limits of Cost-Benefit Analysis, August
2010
37